Fujitsu Insights
Server Virtualization and Private Clouds
Nowadays planning horizons are shorter, revenue streams are uncertain, and you have to be flexible to sur
-vive within your business. Among others your IT infrastructure, which underpins your business, has to be very
agile to meet today’s flexibility needs. This Insight provides some background information on server virtual
-ization and private cloud computing and how Fujitsu can help to realize such projects successful.
Some reasons to discuss these topics with Fujitsu 3
x86 server virtualization 4
What is virtualization? 4
What is x86 server virtualization? 4
The indirect advantage of a virtualization layer 4 Preconditions to the effective use of server virtualization 5 How to manage hypervisors? 5 Format of virtual machines 5 Performance impact of the hypervisor 5
Benefits of server virtualization 6
Can everything be virtualized? 6
Market acceptance of server virtualization 6
Server virtualization and cloud computing 7 Server virtualization, what else? 7
Private Clouds 8
What is ‘cloud’? 8
What is Private Cloud? 8
Preparations for the effective use of Private Clouds 9 How to manage Private Clouds? 9
Benefits of Private Clouds 9
Can everything be brought to the Private Cloud? 10 Performance impact of Private Clouds 10
Market acceptance of Private Clouds 10
Risks of Private Clouds 10
Private Clouds and server virtualization 11 Private Clouds, what else? 11
Microsoft Private Cloud Fast Track 11
Intel Cloud Builder 11
Versatile servers with top performance – FUJITSU Server
PRIMERGY Rack 12
PRIMERGY rack server overview 13
Economies of scale – FUJITSU Server PRIMERGY Blade 14 Cloud eXtension servers – FUJITSU Server PRIMERGY CX 16
Scale-out smart 17
Spend small 17
Integrated server lifecycle management – ServerView Suite 18 Standardized administration of physical and virtual servers 18
Efficient Virtual-IO Management 18
Automating server lifecycle operations 19 Build your Private Cloud infrastructure 19 Ready for self-service provisioning 19
The Flexible Data Safe – FUJITSU Storage ETERNUS DX 20
ETERNUS DX – a central piece of virtualized and cloud data centers 20 ETERNUS DX provides answers 21
What about data protection? 22
Complex environments 22
One-for-all 22
Proven data protection appliance 23
Backup to disk with de-duplication 23
Fujitsu integrated cloud infrastructure solutions 24
Challenges in approaching private cloud infrastructure projects 24
Solution - The integrated systems approach 24
FUJITSU Integrated System Cloud Ready Blocks 25
FUJITSU Dynamic Infrastructures for VMware vCloud 25
SAP-aware IT infrastructure – FlexFrame Orchestrator &
Solutions for SAP HANA® 26
FlexFrame Orchestrator 27
FlexFrame Orchestrator – move towards SaaS 27
FUJITSU Solutions and Services for SAP HANA
What about client virtualization? 28
Virtualization enables centralization 28
Various concepts for different requirements 28
Fujitsu’s Virtual Client Computing approach 29
Fujitsu Thin Clients 30
Scout Enterprise and ELIAS 30
Fujitsu Thin Clients - Supporting eLux® and Windows® Embedded 30
Choice of Operating Systems 31
Benefits of the Fujitsu Thin Client Architecture 31
Fujitsu VDI Bundles 31
Maintenance Services 32
Hardware Maintenance 32
HDD Retention 32
Software Support 33
Benefits of Fujitsu Maintenance services 33
Managed Services 34
Data Center Services 34
Managed Data Center 34
Managed Storage 34
Managed Server 34
Managed Application 35
End User Services 35
Virtual Client Services 35
Classic Client Services 35
Managed Mobile 35
Unified Communications and Collaboration Services 35
Managed Maintenance 35
Financial challenges and benefits 36
Financial benefits 37
Advantages at a glance 37
One Stop Shopping 37
Cloud Services & Solutions by Fujitsu 38
Fujitsu Cloud Infrastructure as a Service 38
Fujitsu Cloud IaaS Trusted Public S5 38
Fujitsu Cloud IaaS Private Hosted 38
Fujitsu Private Cloud Infrastructures 38
A word on security 38
Additional benefits included 39
Fujitsu Cloud Integration Platform 39
Server virtualization or Private Cloud? 40
Assessment services 40 Helping you build your Private Cloud 40
Fujitsu has a very long history in the topic of server virtualization – more than 40 years. It started with virtual machines in our mainframe offering, GlobalServer and Fujitsu’s BS2000, which still are availa-ble. Knowing server virtualization technology and its benefits from the beginning, Fujitsu has also completed thousands of x86 server virtualization projects successfully. Fujitsu also transferred its experi -ence as one of the world’s leading cloud service providers in helping customers to build their cloud environment. This is proven by many current success stories and Fujitsu’s contributions in in private cloud programs from our technology partners like Intel and Microsoft. Also Fujitsu’s products, such as FUJITSU Server PRIMERGY and FUJITSU Storage ETERNUS benefit from the comprehensive experience as ser -vice provider and system integrator. PRIMERGY server and ETERNUS storage systems are optimized for x86 server virtualization and cloud infrastructures – among others by using the latest technologies, certain features and maximum compatibility with market standards – as proven by many benchmark records.
Moreover it is also worth mentioning the integration of server virtualization in Fujitsu solutions and Fujitsu’s offerings that extend virtualization. One example of the use of server virtualization within complete solutions is FUJITSU FlexFrame Orchestrator – an entire solution for SAP environments. FUJITSU ServerView Resource Orches -trator is an example of an extended server virtualization offering. It allows the uniform management of a mixed environment – even if different server virtualization technologies are used or if it is a mix of virtual machines and physical servers. That is complemented by the offering of complete pre-configured and preinstalled ready-to-run converged solutions like Fujitsu Cloud Ready Blocks which, besides its further benefits of a converged infrastructure,
enables a very short time-to-production. Not to forget Fujitsu’s data protection appliance FUJITSU Storage ETERNUS CS, harmonizing and optimizing entire backup and archiving scenarios.
A further essential argument why you should deal with Fujitsu in the area of server virtualization and cloud computing is Fujitsu’s complete service offering. This service offering for all aspects of server virtualization and including any kind of cloud services starts with pre-defined and fixed price assessment services to analyze the current customer situation and to provide customers with re commendations about how to proceed with their IT. Among others, these OPTIMIZATION Services are available for cloud comput -ing and server virtualization. Beside these consult-ing services, ROI/ TCO estimations, design services, delivery services, operation servic -es, product/solution-related servic-es, various managed services and financial services are also available.
Fujitsu’s close cooperation with all market leaders in virtualization and cloud technologies puts us into a position to be the reliable ad-visor of our customers. For some similar technologies we deal with different partners – for example on the server virtualization layer with Citrix, Microsoft, SuSe, Oracle, RedHat and VMware – in order to have an appropriate solution for any customer requirement. Moreover, it should be mentioned that Fujitsu is the only server manu-facturer today that develops and produces its x86 PRIMERGY servers in the heart of Europe. One further argument, especially for global customers, is Fujitsu’s capability to serve customers worldwide with all levels of expertise, products and services. Our follow-the-sun call centers support more than 25 languages.
x86 server virtualization
What is virtualization?
Within the IT industry virtualization is used between today’s standard interfaces to abstract different layers from each other. Because of defined and comprehensive interfaces the dependencies of different IT ‘layers’ (e.g. the server hardware and the connected network in -frastructure) are very strong and often result in inflexibility and low utilization. For example:
When one layer is updated, the other one needs an update, too. When one layer is defined, the upper and lower layer often can’t be used for anything else.
To overcome these strong dependencies an additional layer – the virtualization layer – can be inserted. As today’s IT has many defined layers, the possibility of inserting separation layers between them is also varied. Just to name some of the resulting virtualization technologies: ■I/O virtualization ■Server virtualization ■Network virtualization ■Application virtualization ■Storage virtualization ■Client virtualization
■Operating system virtualization ■...
Many IT installations during the past years often contained more than one of these technologies in order to combine their related benefits and to achieve a state-of-the-art, dynamic IT infrastructure. In terms of this document we are focusing on x86 server virtualization.
What is x86 server virtualization?
As the title indicates, x86 server virtualization describes the virtual -ization layer between x86 architecture-based server hardware and the associated operating system. This virtualization layer is usually called the hypervisor. This hypervisor isolates the x86 server hard -ware from the various operating systems. The server hard-ware sees it as its operating system and provides the known operating systems with the capability to use pre-defined portions of the available x86 hardware resources. Those portions are called virtual machines (VMs) and are fully isolated from each other. They do not even know each other. This enables the sharing of server-hardware capabilities between different operating systems and applications. The core tasks of a hypervisor are:
■Foundation to operate VMs
■Separation of the VMs from each other ■ Efficient distribution of resources between VMs
in line with the administrator’s specifications ■Mapping of machine instructions issued in a VM ■Mapping of I/O jobs
The indirect advantage of a virtualization layer
By adding an additional layer as an interface between the server hardware and the operating system, a further major advantage can be achieved: it is possible to influence their communication. Said in -fluence never means modifying anything, but it enables time shift -ing and – even more important – shar-ing of information. Transferred to server virtualization this means that memory and CPU states of a virtual machine can be copied to another server. The result is one major feature of using hypervisors:
■ Live Migration (aka as VMotion, XenMotion, …) of virtual ma -chines (including operating system and application) without any interruption from one server to another one
This is the fundamental technology used for many further automa-tion features provided within hypervisor management.
Preconditions to the effective use of server virtualization
In addition to the very obvious components of a virtualized infra-structure: hypervisor, its management and appropriate x86 server hardware (optimized server virtualization support, fitting I/O connec -tions) centrally accessible storage is also an elementary component. To make use of the high availability options – e.g. restart of a virtual machine in case of a hardware failure on a different server – the server inevitably needs access to the data of that virtual machine where it should be restarted. So typically storage consolidation has to be done first, before x86 server virtualization is introduced/ex -panded to leverage all the capabilities provided by this technology. This can either be done with central storage systems or by using storage virtualization (e.g. of local discs within the server) which provides similar capabilities.
How to manage hypervisors?
As can be imagined, the hypervisor itself has to be a lean and very efficient piece of software. Therefore, to provide little or no surface for security attacks the hypervisor itself is optimized to support its basic functions and to move all resource-intensive tasks (e.g. management via GUI, automatism, rules, …) to outside manage -ment software. This manage-ment software takes care of the entire management of the hypervisor AND the virtual machines. It allows comprehensive management of the hypervisor and virtual machine capabilities. Every hypervisor vendor provides its own management software, which partly is also able to manage hypervisors/virtual ma -chines from other vendors. But usually you get the best results and access to the full set of features if the management software of the hypervisor vendor is used. To have an indication of pure hypervisor size it should be noted that for example vSphere ESXi from the mar -ket leader VMware has a size of approximately 150 MB.
Format of virtual machines
Due to the fact that the variable hypervisors are based on techni-cally different sources, a virtual machine generated for one specific hypervisor would not be able to be operated on the hypervisor of another vendor. To overcome this problem and to enable appliances (the application ships as part of the virtual machine and not in con-ventional installation format) all major hypervisor vendors support the OVF (Open Virtualization Format). By using this format the respective virtual machine can be executed on different hypervisors without any changes. However, by using this format (standardized by Distributed Management Task Force, Inc. – short DMTF) some specific features provided by just one vendor probably can’t be lever -aged for the VM in OVF because it is standardized.
In the past and also likely in the future this results for some custom -ers in the need to operate more than one hypervisor type in their data center.
Performance impact of the hypervisor
Inserting an additional layer often gives the impression that everything gets complicated and more power is needed. This partly applies also for x86 server virtualization, BUT ■ x86 servers have had a tremendous performance boost for
many years, which is still continuous
■ The average utilization rate of x86 servers without using server virtualization is usually below 15%
■ Today’s x86 servers and their used components are optimized for server virtualization
■ Nearly all hypervisors are stable and performance-optimized – in various ways
This results in the overall message that the typical performance impact of much less than 10%, if real-time operating systems are not used, can be disregarded.
Benefits of server virtualization
Customers can achieve many business benefits by using server virtualization. The technology in principal might not be new, but it has never paid off on such large scale.
First, for most implementers the achievable consolidation ratio was the key motivation to introduce server virtualization within their company. When customers are able to replace 15 (or even more) old servers with one new server that additionally needs less power than one of the old servers, this results in tremendous savings in investment costs (CAPEX) and operating cost (OPEX – energy, space requirement and management).
Additionally, the homogenous virtualization layer allows customers to run old operating systems on new hardware. The example often used is Windows NT, which is already out of any Microsoft support and certification program and which is executed in a virtual machine running on the up-to-date Fujitsu PRIMERGY generation. Any virtual machine can also be implemented on any ‘virtualized’ server hardware, which provides very flexible use of all x86 resources. Many automation functions could be implemented with the option of live migration of workloads from one server to another. For example, power-optimized use of servers while all applications are up and running: if the applications are only rarely used, they can be aggregated on a few servers and the remaining servers could be switched off. When more compute performance is needed, an appropriate number of switched-off servers would be powered-on and the applications in the virtual machines would be live-migrated to the powered-on servers – all of that fully automatically and without any interruption of the application. Similar things apply for hardware maintenance windows: zero downtime.
Software deployments also get much easier: a virtual machine does not need to be ‘active’ – the management of the hypervisor can also update offline machines. Likewise all hypervisor vendors provide convenient tools to run automatic updates of all affected machines or duplicate existing virtual machines or templates. A new service can be up and running with just a few mouse clicks.
Not to forget the very easy scalability. As mentioned above, there are many options to lower the number of powered-on servers. The opposite, e.g. extending a virtualized environment, just needs an additional server with the appropriate hypervisor. Just connect the server to the network and it can be integrated in the overall environment with a few mouse clicks.
All of that is complemented by the integrated and uniform high availability concept that comes with using server virtualization. In the past it was often very complicated to set up high availability solutions, because many of them had application or operating-sys-tem-specific dependencies and were mostly associated with an additional license fee. By using server virtualization all hypervisor vendors enable an integrated, application and operating-system-in-dependent concept: if a server fails or can’t be contacted, all virtual machines can – depending on the predefined settings – automati -cally be restarted on another server. This is uniform for the complete virtualized environment. On top of this, you can add mechanisms which control the applications to extend this general capability to an application-aware feature. To avoid misunderstandings: traditional clusters still work when server virtualization is used.
Can everything be virtualized?
Unfortunately not everything can be virtualized with server virtual-ization. All servers produced today include virtualization-optimized features. But servers bought more than 5 years ago probably do not support all capabilities of a hypervisor and its management. Also certain I/O cards (e.g. dongles, fax cards, special graphic cards, ...) can’t be used with x86 server virtualization. Which components are supported can be seen in the hardware compatibility list (HCL) of the various hypervisor vendors. The number of unsupported and currently available components is getting smaller and smaller.
Market acceptance of server virtualization
According to hypervisor vendors and many analysts more than 70% of all workloads of x86 servers are running in virtual machines today. Moreover, also according to statements from IDC, Gartner and Forrester, the market share of Fujitsu’s hypervisor main partners, Citrix, Microsoft, SuSe, Oracle, RedHat and VMware, together is big
-ger than 99% of all hypervisors used in business. The top three in the ranking with regard to the ‘number of virtual machines running on which hypervisors’ are:
1. VMware 2. Microsoft 3. Citrix
Server virtualization and cloud computing
The capabilities of server virtualization, especially its great benefits with regard to flexibility, uniform SLAs, automation options and scalability, can be perfectly used to deliver the foundation for cloud computing. This does not mean that any kind of cloud computing always relies on server virtualization, but it can be seen as one of the key enabling technologies for it. The picture below describes the movement of customers from virtualization towards cloud comput-ing and the logical steps in between.
Any kind of cloud computing, whether private, hybrid or public clouds, is more than the extensive use of server virtualization. However, server virtualization is often one step towards cloud computing, which needs many more capabilities to be implemented. All hypervisor vendors work on their own cloud stacks to enable customers to build their own cloud solutions. Moreover, they offer integrated tools for uncomplicated access to cloud services from own or publicly available offerings to also leverage the benefits of these cloud offerings.
Server virtualization, what else?
As already stated in previous sections, x86 server virtualization is a mature technology which naturally also expands its functions in associated areas. The main areas today are storage virtualization, network virtualization and cloud management. With cloud man -agement all kinds of cloud computing: private, hybrid and public clouds are addressed. Overall this development results in many nice features, formerly known from separate tools and vendors only, can easily be accessed by additional licensing options of the server virtualization layer and its management. Meanwhile all hypervisor vendors have an offering beyond pure server virtualization and these offerings merge together. This also applies for Fujitsu with offerings that extend virtualization described in further sections of this document.
Private Clouds
What is ‘cloud’?
Fujitsu defines ‘cloud’ as the consumption of infrastructure, an application, an activity or content of which that consumption comprises three distinct attributes:
■Pay per use ■Elastic and scalable ■Self service
Clouds are usually available in four different formats: ■Private Cloud is dedicated for one customer.
■ Community or Trusted cloud is for serving a number of selected customers.
■Public cloud is for (theoretically) every customer. ■Hybrid cloud is any combination of the three others.
Each format could provide services with the three distinct attributes. The most popular services today are:
■IaaS, Infrastructure-as-a-Service: providing server, storage and/or network capacities to its users (e.g. system administrator) ■PaaS, Platform-as-a-Service: providing operating systems and software development tools including the necessary IT infrastructure to its users (e.g. software developer) – mostly used to provide SaaS later on
■SaaS, Software-as-a-Service: providing the ability to use software to its users (e.g. any user needing the software)
■BPaaS, Business-Processes-as-a-Service: providing the use of complete business processes to its users (e.g. logistics manager)
From the user perspective, the technology and the ‘how to’ issues usually associated when a new service is needed are completely taken away. Users of cloud services can fully concentrate on their core business and do not have to deal with daily IT operations because of the standardized and automated processes coming along with cloud computing. Just to give you an example: by using the cloud – independent of whether this a private, public, community or hybrid cloud, but depending on the service offered, see examples above – a catalog with predefined services is the starting point. Access is realized via a self-service portal where the user can apply for the requested service. After the user has found its appropriate service and applied for it, an acceptance/rejection workflow runs automatically. If the request is accepted, the user can access the service according to the description of the catalog. In addition to the functional description of the service, the aforementioned service cat-alog also contains information on the pricing (based on use time), term and time of payments and the quality description of that ser-vice: like availability time a year, recovery time, frequency of backups (if any) and many more. Last mentioned examples are called Service Level Agreements, in short SLA.
Summarized, the user of a cloud no longer has to take care of technology and how to set it up – the user can pick any available service, just use it when it is needed and only has to pay when the service has been used.
What is Private Cloud?
Private Cloud is a cloud environment that only serves one single customer. It is operated in two different ways:
■ Private Cloud on-premises means the operation of your cloud in your data center with your IT infrastructure
■ Private Cloud off-premises means the operation of a dedicated (again, just for you) as-a-service offering in the data center of your cloud provider
As the cloud environment in both cases is dedicated to you and your needs, it is obvious that the key criteria for cloud computing – again: pay per use, elastic and scalable, self service – are difficult to match in full. For example, the benefits of ‘pay per use’, if you have a private cloud on-premises, might not be matched in full because you are still the owner of the entire IT infrastructure and there are investment and operation costs; more or less independent of how much capacity is used. On the other hand you have much better options to take influence on everything in that cloud, including the definition of SLAs and how to handle your data in detail.
In terms of this document, Private Cloud is Private Cloud on-premises. Cloud: Consuming IT as a Utility
Pay per use
Elastically
Scalable
Self
Service
Privatejust for you selectively sharedCommunity shared with anyonePublic
Hybrid – a combination
Preparations for the effective use of Private Clouds
Before transforming your IT infrastructure into a Private Cloud or set-ting it up from scratch, a detailed analysis of the business processes within your company has to be undertaken. This has to include how your individual short, mid and long-term targets can either be achieved by:
■ Extended use of virtualization technologies and associated automation tools
■Private Cloud environment on-premises
■Community or publicly available as-a-Service offerings
Otherwise, you could possibly implement a Private Cloud environment and the achievements do not match your expectations.
How to manage Private Clouds?
The management of Private Clouds needs the same principle architecture as the management of any cloud. In general this is:
The word ‘service’ in the picture above, stands for that kind of service that should be provided by the Private Cloud to its users: infrastruc -ture, platform, software or business processes. All management soft-ware components have to be integrated with each other to achieve a complete cloud solution.
Today several tools are available for Private Cloud computing, e.g. Fujitsu ServerView Resource Orchestrator Cloud Edition, which is described in a separate section of this white paper. In addition and as already mentioned in the section on server virtualization, complete cloud management tools are available from hypervisor providers and also from management software companies. To have the optimal solution for our customers and their needs, we comple-ment our portfolio with these solution stacks. Some of these tools, e.g. VMware vCloud Suite, also allow managing resources from com-munity or public cloud offerings – and thus lead to comprehensive management of your complete hybrid cloud environment.
Benefits of Private Clouds
The three distinct attributes of cloud computing already indicate the benefits of Private Cloud computing. In particular, these are: ■Ease of use
■Cost savings and cost transparency ■Very flexible use of IT resources
Ease of use is achieved by the general structure of a cloud. The necessary and strict multi-tenancy (in Private Clouds these are often different departments of your company), the standardized service level (SLAs) and the self-service portal build the foundation to fit this benefit. It can be even improved by using further automation functions, well defined and prepared (e.g. SLAs) entries in the ser -vice catalog and by the provided ser-vices covering as much as possi-ble of the tasks of each user, e.g. by providing Software-as-a Service instead of providing Infrastructure-as-a-Service for the respective user – depending on the needs of the user.
Lower costs result from the shared infrastructure of a Private Cloud. For example, peak requirements of one department, often responsible for too many installed/assigned infrastructure components, can be absorbed by temporary assignment of free resources (e.g. compute power) when they are not needed by other users/departments. Thus, the average utilization rate is much better than before thanks to sharing the available resources between all users in a still sep-arated (catchword ‘multi-tendency’) way. Moreover, the elements responsible for the ‘ease of use’ pay off again, because the users and administrators as well are freed from day-to-day routines and can focus on their key tasks. All of that is fully transparent through the metering and accounting functions of the cloud management layer and so avoids any fruitless discussions between departments when it comes to covering the IT costs.
Generally, it could be said that as more services of a similar kind are provided by a Private Cloud the bigger the benefits are. But also the inclusion of public or community cloud services to your Private Cloud enhances the aforementioned benefits.
Can everything be brought to the Private Cloud?
Comparable to server virtualization, not every service or task can be operated in a Private Cloud. Again, depending on what kind of service should be provided, there might be major adoptions necessary. Let’s take Software-as-a-Service as an example. If you have the goal to provide SaaS for certain applications, these applications need some capabilities to be provided as-a-Service and also need the appropriate Private Cloud management stack. Operating the application without any adaptations or special management software in an IaaS environment would not provide the requested result. Beside technical issues license costs which blow your budget could also occur. Nevertheless, because Private Cloud environments are accepted in the market (see ‘Market acceptance of Private Clouds’) and also because nearly all established software providers and likewise many startups are extensively pushing this technology, the number of unsupported services is getting lower and lower.
Performance impact of Private Clouds
Obviously a Private Cloud needs an additional management layer, see section ‘How to manage Private Clouds?’ Most components of that management layer are not needed all the time – thus they only need small resources when they are executed. The always-on management software components, e.g. the metering tools, are optimized to have very little impact on daily operations. Today, the technology used to build the various Private Cloud services, whether it is IaaS, PaaS, SaaS or BPaaS, is different. This – combined with various management options – makes it very difficult to disclose exact figures on the performance impact. But the experience we gained with our own cloud service offerings and within our customer projects, where we supported our customers with the building of their Private Cloud, results in the overall message that the typical performance impact can be disregarded – mostly independent of what service is provided.
Market acceptance of Private Clouds
According to various analysts Private Cloud computing will be used – at least for parts of their IT/business – by more than 40% of all enterprises. The clearly most popular service provided by Private Clouds is Infrastructure-as-a-Service; this also applies in productive use. The reason why IaaS is the most accepted service (again, within Private Clouds) is in the available offering for the required manage-ment stacks. The IaaS Private Cloud managemanage-ment stack is today the most complete and commercially available Private Cloud solution. As indicated in the comparable section of server virtualization, where VMware is the market leader, it is mainly used as the basis for IaaS services in Private Clouds. So it is no surprise that VMware is also the market leader in Private Cloud management stacks for IaaS. An ‘extended IaaS’ is often used to provide PaaS via Private Clouds. When it comes to SaaS, mainly application-specific solutions come into play nowadays. The same applies for BPaaS. One example of SaaS Private Cloud solutions is our FlexFrame Orchestrator, which is described in the section ‘SAP-aware IT infrastructure’. Summarized, it could be said that market acceptance for Private Clouds is high. Moreover, Infrastructure-as-a-Service is dominant today, but the other service offerings are evolving very strongly and especially SaaS will, when it has greater maturity and broader-based offerings are available, surpass IaaS in Private Cloud computing.
Risks of Private Clouds
Just as with benefits, which are in part difficult to achieve in full, the commonly discussed risks of cloud computing usually do not apply in full for Private Clouds on-premises. One of the main reasons holding back universal adoption of cloud computing is security. Being based on the Internet does make cloud computing susceptible to cyber attacks, but Private Clouds are operated within your firewalls and that makes them at least as secure as your infrastructure was before. When a company has its data in the cloud, it is not under its direct or physical control and due to the respective, applicable law other governments may have the right to view that data (e.g. in U.S. or with U.S.-based companies as cloud providers). However, Private Cloud on-premises hosts its storage capabilities in your data center, which is fully under your control. So you decide on the accessibility of your data and how it will be protected. We could extend this section by many other points but they key message would not change: most possible risks of public cloud computing are under your full control by running a Private Cloud environment. Therefore, the risk of your Private Cloud is your responsibility and depends on the SLAs you implement for your solution.
Private Clouds and server virtualization
Server virtualization is often used as the key technology in the area of Infrastructure-as-a-Service offerings in particular, but also for Platform-as-a-Service offerings. The described capabilities of server virtualization, enhanced with a serious Private Cloud management stack, are a validated combination that is often used today. It can also be said that the broad acceptance of server virtualization and its automation paved the way for cloud computing, especially with regard to IaaS. For the remaining cloud services server virtualization can be seen as enabling technology, but not as ‘must have’. SaaS and BPaaS can be built without using server virtualization, but you can also use IaaS as the enabler for SaaS.
Private Clouds, what else?
The natural extension of Private Cloud computing is its expansion to include cloud services from cloud service providers. This inclusion of other services is indicated several times in previous sections. For example, you can extend your compute power with Fujitsu’s Infrastructure-as-a-Service offering when you need more capacity than your Private Cloud can provide. Or you use our Backup-as-a-Service offering; originally developed to support SMBs to fulfill certain compliance regulations. The inclusion of public or community cloud offerings is the next step after having a Private Cloud on-premises implemented.
Microsoft Private Cloud Fast Track
This program initiated by our technology partner Microsoft describes in detail how exactly a Private Cloud infrastructure has to look to operate Microsoft applications best. Fujitsu, as one of the first five members, joined this initiative directly with its announcement in November 2010. According to the strict requirements of the program and with the claim to increase the value provided by reference archi-tectures, we jointly defined our reference architecture with
Microsoft. Fujitsu’s configurations, except the special versions for SMBs, consist of Fujitsu PRIMERGY Servers, Fujitsu ETERNUS Storage, Fujitsu ServerView Suite, network components and the Microsoft management software stack.
Intel Cloud Builder
As opposed to Microsoft Private Cloud Fast Track, this program initiated by our technology partner Intel focuses on any kind of reference architecture to build cloud infrastructures. But Fujitsu as one key contributor focuses on reference architectures for building Private Clouds on-premises, which are all different to our contri-bution at Microsoft Private Cloud Fast Track. Beside that we build solutions with different Private Cloud management stacks, Fujitsu’s own solution using ServerView Suite and VMware or Microsoft tools. We also integrate various elements of our cloud-optimized product offering (e.g. Fujitsu PRIMERGY BX, CX and RX – described in further sections of this white paper) in our reference architectures. In coop-eration with Intel we are continuously setting up new configurations for different customer needs. We use the Intel Cloud Builder Program to provide our customers with publicly available guidance including a very detailed description of how to build their Private Cloud.
A high-quality IT infrastructure is the foundation of the productivity and continuity that are essential for the success of any business. That’s why Fujitsu is so focused on quality assurance and engineer-ing. It ensures the high availability of our rack servers, associated with tailor-made maintenance services to support stringent SLA requirements. This results in extremely low annual failure rates which are comparable to those of UNIX servers, and protect your investment over the product lifecycle.
IT performance is another key element for business success. Fujitsu PRIMERGY rack servers achieve world records and best-in-class results in different disciplines: Nearly all Fujitsu PRIMERGY rack servers lead Intel-based server rankings in CPU and Java performance benchmark results. This is also reflected in VMware virtualization environments, where Fujitsu PRIMERGY rack servers have achieved leading perfor -mance positions.
To stay competitive, reducing costs is essential for all companies. Fujitsu PRIMERGY rack servers contribute to this goal by delivering efficient servers and management tools to increase business pro -ductivity, while simultaneously reducing operational costs through energy efficiency that is up to 20% better than competitive products, virtualization and simplified management. Fujitsu PRIMERGY rack servers lead the market in energy efficiency with unique ‘Cool-safe™’ technology. In fact, Fujitsus Cool-safe™ Advanced Thermal Design enables support of environmental temperatures of up to 40 °C. The temperature in a data center can thus be increased by a few degrees without any problems. Each additional degree means approx. 5-6 % less energy costs for air-conditioning.
Versatile servers with top performance
FUJITSU Server PRIMERGY Rack
Virtualized environments as well as Private Cloud infrastructure require a reliable and powerful IT infrastructure that supports productivity while being efficient and flexible to meet the demand of tomorrow. As the operational heart of every data center the servers should be optimized to run their workloads. Fujitsu PRIMERGY rack servers deliver the flexibility and scalability to achieve this while also lowering costs and improving productivity. In fact, Fujitsu PRIMERGY rack servers are the first servers to use 1 Watt or less per transaction, while competitive products consume five times more power. Combined with best-in-class price/perfor -mance results this contributes to favorable lifecycle costs. They are also integrated into the Fujitsu ServerView suite (own section of this white paper), which enables efficient, intelligent management that simplifies administration and improves responsiveness to quickly adopt new business changes. Finally, the Fujitsu supply chain pro-vides the customization required to deliver a tailor-made, tested and preassembled configuration to your site.Fujitsu ServerView – basic components shipped with every PRIMERGY server – enables local and remote management of PRIMERGY-based infrastructures. The software accelerates server installations, while centralizing monitoring, control and integration.
By eliminating complexities, server management becomes more efficient. Intelligent system management solutions enable you to simultaneously improve user productivity and simplify daily admin-istrator’s tasks.
Flexibility is also a key value in our Fujitsu PRIMERGY rack server supply chain. MADE4YOU provides customized production and logistics services. These services can pre-load software, integrate third-party components, provide customized consignment stocks and specify rollout arrangements at all stages of the value chain. This results in faster time-to-operation. Our built-to-order production standards ensure that your customized PRIMERGY configurations are manufactured, tested and delivered to completely match the functionality defined in the order. This even includes the shipment of fully mounted rack server configurations, so there is no need for your IT team to re-assemble servers or components. Cost and time-to-operation are significantly reduced as a result.
PRIMERGY rack server overview
Fujitsu PRIMERGY rack servers provide a reliable foundation for any LOB application and data center solution. They also offer functionality tailored to your requirements:
■ RX100: Entry-level mono-socket system for data centers, providing high availability, scalability and low TCO – in a 1U chassis. ■ RX200: Dual-socket system providing top performance for virtualized environments, high expandability and leading efficiency – all in 1U space saving housing.
■RX2520: 2U, efficient and scalable dual socket rack server, which offers an especially well balanced price / performance ratio and huge storage capacity in a 2U housing.
■ RX300: 2U, dual socket rack server focusing on top performance, high versatility and scalability to support all types of business applications and consolidation.
■ RX350: 4U, dual socket rack server offering maximum perfor -mance, expandability and availability which makes it ideal for server, storage and client virtualization.
■RX500: 4U quad socket server leverage the price/performance of dual socket servers into the four socket area – thus makes it ideal for large virtualization
■ RX600: This quad-socket server is a reliable server for critical company IT services such as databases or large-scale virtual-ization, providing high availability and consistently efficient IT production
Economies of scale
FUJITSU Server PRIMERGY Blade
The whole world today is thus connected via networks whichgenerate a huge amount of data and almost everybody is able to access information anytime and anywhere. Thus lots of companies are placing greater emphasis on the operation of their IT environ-ments, not only to gain a competitive advantage in the market but also because IT services and applications have become a critical element in day-to-day business and workforce productivity. Due to this increasing demand, server rooms have to host a large number of often different hardware platforms. This makes an IT environment extremely complex and difficult to manage, leading to high costs due to personnel-intensive administration on the one hand and excessively required space on the other hand. For the operation of these systems the energy required for power and cooling emerges as an increasing share of the costs compared to the pure acquisition costs. Fujitsu PRIMERGY Blade Servers are a perfect supplement for x86 server virtualization and provide additional advantages, such as improved manageability, greater availability and operational efficiency through a shared infrastructure and virtual input/output. Fujitsu PRIMERGY Blade Servers offer the ideal balance of perfor-mance with scalability, and are the standard for a wide variety of en-terprise business and HPC scale-out applications for small, medium, and enterprise data centers. They are the perfect platform to build a private cloud or converged infrastructure, designed to maximize time and efforts.
The blade enclosure provides all the power, cooling, and I/O infra -structure needed to support a modular infra-structure, interconnects and storage components today and throughout the next several years. PRIMERGY Blade Servers can be used to meet increasing com-pute performance requirements without suffering from an explosion of costs and increased administration complexity.
Fujitsu offers versatile enclosures to match the unique needs of large or small IT environments.
The PRIMERGY BX900 S2 is a complete dynamic server infrastructure in one single blade chassis. The blade server can be dynamically modified for various IT requirements: it reduces costs and makes IT more flexible. The PRIMERGY BX900 has space for up to 18 server blades in a 10U chassis.
The Fujitsu PRIMERGY BX400 is a fully-featured blade system built from the ground up to be user-friendly and versatile, ideally suited for midmarket organizations and branch offices. It provides an optimum combination of up to eight server or storage blades in a small enclosure as a rack version with 6 height units or as a separate floor-stand version.
As compute nodes, Fujitsu offers two different dual-socket server blades which can be used in both chassis.
The Fujitsu Server PRIMERGY BX920 S4 is equipped with the cost-op-timized Intel® Xeon® E5-2400 v2 processor product family and offers the ideal balance between performance and cost efficiency. The Fujitsu Server PRIMERGY BX924 S4 sets new standards in the area of versatility and scalability of dual-socket server blades and thus re -inforces its position as a high-end server. With two CPUs of the new Intel® Xeon® processor E5-2600 v2 product family, 24 DIMM slots supporting up to 1,536 GB of memory and an integrated 10 Gbit/s Ethernet Universal Converged Network Adapter (UCNA), the server blade is ideally suited for extensive virtualization and consolidation projects.
They can be equipped with the latest Intel® Xeon® processor E5 product family, provide scalable memory configurations with up This UCNA enables Ethernet on its own and as a combination of Ethernet and Fibre Channel over Ethernet (FCoE) protocols and can be used to partition the 10 Gigabit Ethernet adapter into up to eight physical functions, so that the individual bandwidths of the channels can be adapted to the requirements of each application. Thus offers consid-erable benefits for virtualized environments. Traditional virtualized environments often require between six and eight connections per server in order to migrate virtual machines and for the manage-ment, backup and storage connections. The converged network adapter can replace traditional Host Bus Adapters and Ethernet Network Interface Cards, thus notably reducing costs.
Both server virtualization and cloud computing deal with large-scale use scenarios, which will inevitably grow to encompass broader service requirements in order to map individual business dynamics. Therefore, the server infrastructure has a very important impact, especially if much computing power is needed. On the other hand, traditional data center facilities do not easily keep pace with those massive compute capacity demands, since they have to master additional challenges:
■Increased power envelopes and cooling requirements for those servers and related infrastructure
■Limitation in data center floor space, and the requirement for more computing power in a limited amount of space ■Limited budgets enforcing minimized initial purchase costs and
subsequent maintenance costs and operational spending ■The need for more efficient manageability and less complexity
in the operation of massive scale-out compute resources The Fujitsu PRIMERGY CX Cloud eXtension servers are Fujitsu’s platform for cloud computing, high performance computing (HPC), service provider and any large scale-out computing server farm. They focus on providing large data centers with massive scaling power for x86 servers, while at the same time delivering new data center economics for server density, energy consumption, heat optimization and lower overall operational costs. The Fujitsu PRIMERGY CX server family has everything customers need to achieve a new dimension of cost-effectiveness with a cloud data center. Due to their innovative scale-out design, the Fujitsu PRIMERGY CX servers require less space per rack and less energy & heat per server as compared to conventional rack servers. They pay dividends in significantly lowered operating expenses, easier installation and more aggregated computing power in the data center at lower all around investment.
The Fujitsu PRIMERGY CX Cloud eXtension servers are targeting large scale-out server deployments.
Cloud eXtension servers
Scale-out smart
FUJITSU Server PRIMERGY CX400 multi-node server provides a con-densed form factor, enabling you to scale-out smart by packaging 4 independent dual-socket server nodes and up to 24 local storage drives into a consolidated 2U rack enclosure. Fujitsu PRIMERGY CX400 fits smartly into existing data center racks, by providing full compliance to conventional front-to-back airflow, standard 19“ hous -ing of the CX400 chassis and rear connectivity for communication. In this way, team play with an existing rack server infrastructure is easy to achieve. Its ‘4 in 2U’ granularity results in half of the rack space used, lower hardware investment and consolidated efficiency in management for the same performance as compared to stand-ard rack servers. The result is significantly higher performance per standard 19“ rack.
Fujitsu’s PRIMERGY CX400 uses shared redundant fans and power supplies. The energy consumption is significantly minimized while providing enhanced availability and higher effectiveness of manage- ment with hot-plug server nodes, power supplies and local storage drives. Cost-effectiveness is granted for Fujitsu PRIMERGY CX400 by conventional front-to-back cooling, rear-side external connectivi-ty, installation and operation in existing data center rack and air conditioning infrastructures. The server nodes can be individually configured and independently serviced. Additionally a choice of PRIMERGY CX dual-socket server nodes with the latest Intel Xeon pro- cessor E5 family without (1U height) and with (2U height) GPGPU (General Purpose Graphics Processing Unit) and co-processor options enable great flexibility to match different solution stacks with Fujit -su’s PRIMERGY CX400.
Spend small
The minimalistic design of Fujitsu PRIMERGY CX reflects the fact that scale-out applications stacks use software functions in the solutions architecture to care for failover and redundancy. This means that hardware-based redundancies can be replaced by those software functions, thus reducing the number of hardware components re-quired with each server unit. And that in turn pays dividends stream-lined CX server units, cutting down procurement costs compared with conventional rack server configurations.
The space-saving design also ensures that extensions in the data center do not require additional square footage. The result is savings in the cost and work of providing the physical infrastructure. All this has a positive impact on your IT budget and your capital expenses: Depending on the configuration, you can slash these costs by up to 20% with the Fujitsu PRIMERGY CX servers, compared with tradition-al rack servers.
Integrated server lifecycle management
ServerView Suite
Companies today expect considerably more from their IT depart -ments than just the safeguarding of computer operations: Irrespective of the economic situation the pressure to continue to provide services, which are equal to or even better than the previous ones, is on the increase while budgets are frequently stagnating. To ensure that this task does not become insurmount -able Fujitsu has been continuously revising system management functionality in order to meet these increased demands. The focus is traditionally on those functions, which ensure fail-safe and flex -ible operation of servers in addition to the constant availability of business-critical data and which also facilitate their consolidation. Our key attention was recently placed on integrating technologies for the virtualization, dynamization and automation of IT landscapes up to their integration in cross-location and corporation solution concepts, e.g. cloud computing.
ServerView Suite is the management tool from Fujitsu used for the administration of industry-standard servers. Its focus is traditionally on those functions, which ensure fail-safe and flexible operation of servers in addition to the constant availability of business-critical data and which also facilitate their consolidation. Fujitsu’s ServerView Suite offers comprehensive administration covering the overall lifecycle of individual computers as well as entire machine parks and which combines decisive cost benefits with maximum performance and user-friendliness.
Although existing IT capacities are better used thanks to virtualiza -tion and a medium to long-term reduc-tion can even be achieved with them, administrative spending grows compared to convention-al infrastructures with their clear-cut convention-allocation of servers and appli-cations. Therefore, companies need a management platform, which
combines a consistent view of the IT landscape with comprehensive functions that are centrally managed from a single point of control. Fujitsu ServerView Suite is one such platform which is also easy to understand and operate. Functionally, it is divided into the five subsections: Deploy, Control, Dynamize, Maintain and Integrate. It is designed for use in so-called Open Systems environments, i.e. on industry-standard servers which run under the operating systems Windows and Linux (SuSE and Red Hat) or are used as hosts for VMs that use a hypervisor like VMware ESXi, Citrix XenServer or Microsoft Hyper-V.
Standardized administration of physical and virtual servers
As one of the core components in the subsection ‘Control’ Fujitsu’s ServerView Operations Manager (SV OM) supports the standardized administration of all physical and virtual servers in the network. This enables ‘control in just a few steps’ and at the same time ensures increased system availability. A well arranged management console that gives administrators a clear overview of the states of physical servers, VM hosts and virtual machines. It can be used to perform all the important basic operations (starting/stopping, inter -rupting/resuming and restarting servers). Furthermore, it shows all the configuration parameters in full, thus creating at the same time the basis for detailed performance monitoring and coherent event management.
The subsection ‘Dynamize’ combines a series of special system man-agement tools, which above all permit dynamic and efficient use of physical and virtual resources. They differ from the prevalent tools in SV OM by the fact that they also have comprehensive functions for the automated provision and administration of IT infrastructure resources. This includes the following products:
■Fujitsu ServerView Virtual-IO Manager (SV VIOM)
■Fujitsu ServerView Resource Orchestrator Virtual Edition (ROR VE) ■Fujitsu ServerView Resource Orchestrator Cloud Edition (ROR CE) Efficient Virtual-IO Management
The aim of Fujitsu ServerView Virtual-IO Manager (SV VIOM) is to ensure as smooth and efficient as possible communication of the PRIMERGY blade and rack server with clients and storage arrays. For this purpose the module allocates previously defined virtual network addresses to the individual servers and also determines via which interfaces they are to establish connection to the LAN or SAN and how they are to be booted. To enable this to work the SV VIOM saves all the specific information about the computers (e.g. MAC addresses and WWNs) in a separate, hardware- independent server profile, which is kept in a central repository outside the server.
As a result, individual applications can be started without any further configuration effort by the administrator allocating a specific profile to a server and by using the address previously defined for this purpose. Ultimately, the SV VIOM creates a separate domain for server management, thus separating it completely from LAN or SAN management. Any conflicts and overlapping between server, LAN and storage management are ruled out from now on, because through using constant virtual addresses there are no changes in LAN and SAN management.
Automating server lifecycle operations
Fujitsu ServerView Resource Orchestrator Virtual Edition (ROR VE) is a system management tool that delivers integrated administration of physical and virtual servers. By simplifying and automating server lifecycle operations across both physical and virtual environments, IT organizations can increase the efficiency of their day-to-day server operations. Time for deploying or exchanging servers can be reduced by up to 90%.
Moreover, ROR VE enables cost-efficient N+1 high availability. IT organizations can now protect more servers without paying a pre-mium for dedicated high-availability tools like cluster software. By assigning one or more spare servers to a pool of production servers, it is possible to automatically failover any production servers to the spare server in the event of hardware or operating system failures. Business applications can be resumed without any administrator intervention. Compared to manual recovery processes, server recov-ery time is reduced significantly, thus resulting in faster responses to server failures. This applies to both physical and virtual server environments.
Build your Private Cloud infrastructure
ServerView Resource Orchestrator Cloud Edition (ROR CE) delivers all the functions needed to manage a private cloud infrastructure. The software is specifically designed to set up an Infrastructure as a Ser -vice (IaaS) environment with shared resource usage. ROR CE delivers dynamic resource management and cloud service
management enabling self-service provisioning for simple and speedy IT delivery. Physical and virtual servers – independent of the hypervisor or a mix thereof (e.g. VMware vSphere, Microsoft Hyper-V, RedHat KVM or Oracle VM) are supported – storage and networks, are automatically provisioned in minutes, at a fraction of the cost of traditional provisioning methods. In addition, ROR CE guaran -tees the availability of IT infrastructures by providing a number of high-availability options at various levels. From the protection of sin-gle physical and virtual servers, to recovery support for a complete blade chassis or storage system failure, to disaster recovery in the case of complete site failures.
Ready for self-service provisioning
ROR CE fundamentally changes the way IT resources are provisioned and the way IT users get access to these resources. IT users are provided with a service catalog of predefined infrastructure tem -plates which include logical descriptions of single virtual or physical servers, plus multi-server configurations. Available on a web portal, this service catalog enables IT users to easily select and request their IT resources.
Consider a sample usage scenario where an IT administrator requires IT resources for a web service. Instead of directly approaching the IT department, he just opens the service catalog on the self-service portal, selects a multi-server platform and starts the provisioning request. Upon approval of his request, the complete provisioning process runs unattended without any administrative intervention. ROR CE takes the required resources from the resource pools and au -tomatically builds a ready- to-use multi-server configuration consist -ing of virtual and physical servers, storage and network connectivity. Constantly optimize infrastructure
In addition, the self-service portal allows access to advanced system management capabilities providing tools for central monitoring of IT resources, optimizing and planning resource pool capacity, and for gaining access to accounting information.
Resource
Admin
Infrastructure
Automatic
build
The Flexible Data Safe
FUJITSU Storage ETERNUS DX
Data has been growing at an annual rate of about 60% for years and is increasing at the same rate. The demand for virtualization, for server consolidation and cloud computing is rapidly increasing. This is why many storage administrators are familiar with the problem of overload caused by the continual rapid growth in data volume. New applications and an increasing amount of non-struc- tured informa-tion and data analyses mean that many IT systems have reached their capacity limits. The I/O requirements of today’s applications have also become more demandingThe market will require storage and servers to have higher perfor- mance, higher reliability, greater functionality, and increased cost effectiveness. Many already described key functions like Live Mi- gration, High Availability, Dynamic Resource Scheduling or Disaster Recovery mechanisms, just to name some, cannot happen without a central storage system, which is the center piece of any virtual machine farm and Private Cloud infrastructure.
Fujitsu has redefined the standards of enterprise storage with the in -troduction of the latest generation of Fujitsu ETERNUS DX systems in 2013. Combining leading performance architecture with automated quality of service management the ETERNUS DX series aligns stor-age resources with business priorities, thus enabling higher system utilization and delivering increased system consolidation capabilities and a faster ROI. Unified scalable entry-level and midrange systems, a seamless family concept enabling system upgrades and ETERNUS SF, the unified management suite across the product line, reduce op -erational and migration costs. ETERNUS SF provides enterprise-class functionalities in the entry and midrange class and allows flexible disaster recovery concepts for the different model sizes, thus de-creasing investment costs.
Whereas traditional vendors offer different and discrete storage architectures in the entry-level, midrange and high-end segment, the ETERNUS DX approach differs with one uniform system design and uniform system management over the complete product line in all system segments. The only model-specific component is the storage control unit; all other components like disk shelves and management software products are the same for all models. This allows upgrading of smaller systems into larger ones
by just exchanging the control unit. Administrators who know how to operate one ETERNUS DX model can run any other model in the same fashion. Customers and partners benefit from this architectu- ral principle through less costs of growth and reduced operational effort.
ETERNUS DX – a central piece of virtualized and cloud data centers
Storage operations, especially in highly virtualized environments, generate a huge workload in data center administration. This is where Fujitsu ETERNUS SF management software helps with its standardized storage management. If you know how to operate one storage device in the family, you are able to run any of them! Innovative advanced functions help to increase operational efficien- cy. For example, Thin Provisioning means that the administrator can assign virtual capacity to applications so that the real, existing capacity can be pooled. There is no need to invest up front in disk capacity, which is just reserved for future demands in specific cases and is not used productively in the short term.
Fujitsu ETERNUS SF management software also offers close integra- tion with server virtualization offerings, such as VMware. It enables storage systems and servers to work together with onsite failovers during disasters, and it enables servers to offload tasks to the stor- age system, for example provisioning, cloning or the replication of virtual machines.
Thanks to a wide range of guard and protection functions, the system family is like a bank safe for the reliable and uninterrupted management of enterprise data. Fujitsu’s ETERNUS DX family sup-ports uninterrupted operations at all levels – from the physical sys-tems to the implementation of flexible disaster recovery concepts. For example, ETERNUS DX integrates perfectly with VMware’s Site Re -covery Manager (SRM) solution. SRM simplifies and automates the key elements of disaster recovery: setting up disaster recovery plans, testing those plans, executing failover when a data center disaster occurs, and failing back to the primary data center.
Simplifies and automates disaster recovery workflows: ■Setup, testing, failover
■Turns manual recovery runbooks into automated recovery plans ■Provides central management of recovery plans from the VMware
vSphere Client
From a central management station, the vCenter, all actions can be monitored and managed from one point end-to-end. They can calmly prepare the emergency plan, test it and improve it where necessary without any impact on the running operations. All actions are logged and are available for an audit.
One key challenge for each storage vendor is to support virtualiza -tion and cloud as best as possible. It therefore makes sense to move storage-intensive tasks from the servers executing the applications and perform them directly in the storage system. To enable this, hypervisor and cloud management vendors are delivering APIs to allow greater interaction with storage system.
Just one example: in an effort to simplify the interaction and to make it more efficient, VMware developed the vStorage APIs for Array Integration – in short VAAI.
This API separates the tasks of the hypervisor and its storage devic -es, enabling each to focus on what it does best; virtualization-relat-ed tasks for the hypervisor and storage-relatvirtualization-relat-ed tasks for the storage arrays.
With VAAI, storage array vendors can directly integrate their storage hardware and applications with vSphere.
VAAI enables certain storage tasks, such as cloning, to be offloaded to the storage array, which can complete them more efficiently than the host can. Rather than use host resources to perform the work – that was required prior to VAAI – the host can simply pass on the task to the storage device.
ETERNUS DX provides answers
Within our seamless ETERNUS DX family concept we offer enterprise functionality right from the start. Improving storage utilization rates by implementing the Thin Provisioning feature is becoming a critical success factor for any organization. This feature allows you to break the traditional link between allocated and purchased capacity, thus by starting with a low investment you can operate storage systems more efficiently.
The support of standard interfaces and APIs (e.g. VAAI, VASA, …) for virtualized IT infrastructures becomes an integral part of ETERNUS DX disk storage systems. Since hypervisors are built to share storage resources between clustered servers, a better mechanism was need-ed to lock and unlock lockneed-ed storage. This lneed-ed to the creation of the VAAI primitive known as Hardware Assisted Locking or Atomic Test and Set, which drastically accelerates I/O operations when resources are shared. This type of hypervisor-driven storage functionality is likely to continue to grow in importance in the future, with storage arrays adapting to the demands of virtual data centers.
End users have been very enthusiastic about the impact of this sort of array integration with the hypervisor. It allows them to take better advantage of the advanced features of their storage arrays, and results in dramatic performance improvements.
But growing virtualization deployment also increases storage workloads. The value of the new ETERNUS DX series for virtualization operations is manifold:
■The new ETERNUS DX systems offer 5 times more I/O performance which means that 5 times more VMs can be stored on an ETER-NUS DX system.
■ Stable response times for VMware & Hyper-V during high storage workloads with Quality-of-Service and I/O prioritization function -ality
■High level of VMware / Hyper-V integration
■Thin Provisioning and dedup increase capacity utilization ■Functionality for simple backup and restore of VMs
■Infrastructure solutions for server virtualization reduce time to production
ETERNUS DX is the right answer, it is a highly cost-effective ‘all in one’ solution that meets the full range of today’s storage needs while enabling rapid scaling or reconfiguration to meet the needs of tomorrow.
What about data protection?
We have already mentioned the need to have central, accessible storage as an elementary component of server virtualization and Private Clouds. Therefore, storage consolidation has to be done first to leverage all the capabilities of server virtualization and Private Cloud computing. Even in highly available server configurations it is inevitable that the applications have access to ‘their’ data at any time.
Your data are one of the most valuable assets of your company and storage consolidation on its own may be not enough to protect this essential asset. This is why you also have to take care of data pro -tection. So let’s spend some time on data protection with an unified data protection appliance approach on the one hand and with a pure backup to disk with deduplication approach on the other.
Complex environments
The data protection infrastructures that have evolved in data centers have led to complex IT environments that are difficult to operate and maintain, thus making them error prone. This accounts for the heterogeneous data protection landscapes having various server architectures, virtualization layers, diverse backup software solutions, and tape libraries based on very divergent technologies. Most data backup scenarios today resemble a collection of individual islands that have little or no synergies between them. It comes as no surprise that backup process management and infrastructure maintenance are extremely complex and expensive. Administrators just try to keep things running and avoid making any changes for as long as possible. They know that as soon as one change is made, a series of other changes are generally needed at other points in the infrastructure. For example, if the use of a new generation of tape drives is planned, administrators have to be sure that these will function correctly with the existing HW infrastructure and software, and they need to know if a migration of the tapes will be required. What’s more, data backups on hard disks and on tape are generally performed separately. This is not only inefficient from an operational point of view, it also hinders the definition of flexible SLAs (service level agreements) for data protection. Ideally, users should always have the option of deciding when to use disks, deduplication, tape, or even a mix of both in order to keep the speed of backup and restore in balance with the costs of media.
One-for-all
FUJITSU Storage ETERNUS CS8000 radically simplifies data protection in heterogeneous IT environments supporting the consolidation of backup systems on a large scale, and it offers an easy way to provide disaster resilience for backup and archiving.
It sets a virtualization layer between servers and backup target sys -tems. From the server viewpoint it is one huge virtual tape library. The servers write data on the hard disks of the ETERNUS CS8000. From there Fujitsu ETERNUS CS8000 automatically manages all the data.
It can be decided at any moment in time whether the data remain on disks, are written to one or several tape libraries, or are availa -ble on disks and tape. Alternatively the ETERNUS CS8000 internal disk storage pool can be used as a final disk target with integrated enterprise-class deduplication. The data protection appliance takes care of the complete process between servers and backup target systems. The result is a real and automatic ILM (Information Lifecy-cle Management). This supports highly flexible SLAs because users can decide at any time which data should be kept on disks, which on one or several tape systems, or on tape and disk, or on dedupli -cation disk. Therefore, the different requirements of business units for backup/restore and archiving/retrieval services in terms of costs, capacity, and performance can be satisfied with just one solution. Fujitsu’s ETERNUS CS8000 offers a unique and comprehensive connectivity which enables enterprises to simplify the structure and the operation of heterogeneous environments, like mainframe, IBM system i and open systems.
Fujitsu’s ETERNUS CS8000 features universal tape interfaces, making it compatible with all market-leading libraries. It supports all com -mercially available backup applications so that it can be integrated into existing IT environments. This supports the comprehensive consolidation of backup systems and offers maximum investment protection. In addition to that, ETERNUS CS8000 facilitates the HSM (Hierarchical Storage Management) functionality to support archive functions like long-term archiving, and also archiving for compliant purposes (KPMG-certified) within the same appliance.
The modular grid architecture of ETERNUS CS8000 enables seamless scalability and provides enormous headroom for growth. As the grid architecture can be spanned over two data centers, it supports the implementation of almost any conceivable disaster recovery solution for backup and archiving data.