MERIT TS

• Server Consolidation
• The MERIT TS Solution
• Compelling Advantages of the Consolidated Approach
• Why MERIT TS

Over the years, servers have proliferated in departments across organisations, often housing multiple applications and storing specific data relating to these. In most IT environments, departmental servers are typically used to house a number of functions including file, print, database, web, and messaging applications and data all lumped together on one unit, and, in a few instances, servers are used to house standalone applications. But the trend is moving to server consolidation, where servers are centrally controlled and managed. The problems associated with the distributed server approach: Managing these environments is cumbersome and chaotic. The distributed approach eventually results in vast under-use of storage and processing power in some departments and overloading in others. As a result, industry analysts estimate that on average only 30% of the capacity of distributed servers is utilised. The distributed model makes it difficult to upgrade any one element without affecting all the others and it is difficult to share server resources between one department and another because this would be hard to control and manage. When a departmental server becomes overloaded a new server is typically purchased, because it is difficult to find compatible higher-capacity disk drives or higher performance processors to upgrade older models. Similarly, new servers are purchased to accommodate new applications. Critical data is housed on the same departmental servers as non-critical data, which are often located in an uncontrolled environment. It is difficult to ensure the physical security of departmental servers because it is easy for someone to walk out with one of these devices. In some cases, standard PC workstations are used as servers and these are often not attached to an uninterruptible power supply, creating potential reliability problems. Vulnerability to viruses and the task of carrying out software upgrades is also multiplied across a large number of servers and data backups and disaster recovery becomes a nightmare. It is cumbersome and time consuming to support departmental servers scattered across an organisation. There is no control over standards when it comes to different operating system and application versions being deployed on different servers. Data is typically backed up to tape devices, which are directly attached to each departmental server, making recovery cumbersome, and the responsibility of storing the backups is typically handled at a departmental level. As a result, backups are cumbersome to control and manage and tapes are difficult to locate and retrieve. Data recovery is limited to restoring from tapes, it takes hours to restore the data and there is a risk that the tape backups will not work.

The operating system of each server is loaded on internal mirrored disks so that if one disk fails, processing can revert to the mirrored copy with immediate effect. Special software can be deployed that allows automatic duplication of operating system profiles - which define elements such as security and user access - whenever changes occur and these can be copied back in case of system failure. This saves hours of time restoring and reinstating profiles and removes the element of human error. The ideal solution to these problems is to gradually consolidate all these servers by grouping together applications that can coexist on one large consolidated server. For example, file, print, databases, web applications and messaging applications might be grouped together. Where appropriate, a group of applications could be housed on a rack-mounted device, using blade technology, whereby multiple blades can host single or multiple applications, because each blade is in itself a server. In other cases, a modular approach can be used, whereby a number of self-contained storage network-enabled server units are rack-mounted in a cabinet and additional modules can be added as and when required. The operating system of each server is loaded on internal mirrored disks so that if one disk fails, processing can revert to the mirrored copy with immediate effect. Special software can be deployed that allows automatic duplication of operating system profiles - which define elements such as security and user access - whenever changes occur and these can be copied back in case of system failure. This saves hours of time restoring and reinstating profiles and removes the element of human error. Blade technology may be more appropriate for transaction intensive systems such as web applications, and modular technologies used for more storage intensive applications. But, in both the blade and the modular scenarios, the server units would be attached directly to the storage network and storage capacity would be dynamically allocated to each server. This makes server upgrading a lot easier. File and print functionality can be consolidated onto a NAS (network attached storage) device. The operating system of the network NAS is tuned specifically for file and print operations and is equipped with a high speed storage network port. From a backup perspective, once all the data is pooled, it can be replicated to create an on-line, real-time, mirror image of the operational data, which is stored separately on the storage device. The mirrored data can then be taken off-line, at any time, and the point-in-time copy backed up to tape, which would ideally be stored in a tape library resource that is attached to the storage network. A further good option is to create a duplicate off-site business continuance environment, where a second mirror image of the data can be copied to disk and stored in a pooled resource. Once the backup process is completed, the real-time status of the local mirrored data storage resource would be re-instated. Automatic synchronisation would then take place between the mirrored data and the operational storage pool, updating any changes that have happened in the interim. The off-site snapshot of the local stored data would be copied across on-line, typically once every 24-hours or more frequently. In either case, the local and remote mirrored copies of the data would, in turn, be periodically backed up to tape and stored in the respective tape libraries. Consolidation could be undertaken in a phased manner whereby groups of servers are consolidated as they depreciate and are due for replacement or upgrade. A consistent set of designs can be implemented to cater for all types of server environments, including high availability and disaster recovery requirements. This will drive down costs by shortening the design phase of new implementations and will reduce risk by using pre-tested concepts.

Once the servers are consolidated into a few large units, fewer operating systems need to be supported and maintained, reducing the hassle associated with software version upgrades and in the case of a virus attack there are less servers on which to implement fixes. In addition, when multiple applications are centralised it introduces disciplines and controls to the server level that were previously only found in the mainframe environment. At the same time new servers and new applications can be easily introduced into the controlled and disciplined server consolidation environment in order to adapt to new business trends. Backups are also easier to control and manage because they take place on the storage area network and not at the server level and the backup process does not therefore involve the LAN. Point-in-time copies of data can be initiated and copied to another storage device at any time, by freezing the mirrored version of the data for a few seconds and replicating it. The copied data can then be used for simulations, reporting and data warehousing purposes. In this case, because the data is replicated between storage devices in a time-stamped manner and at a block level, the integrity of the data is guaranteed, and block copying is a lot faster than other methods. Once the mirror is reinstated it is resynchronised with the operational data and the changes that have happened in the interim are updated. Data replication can be carried out at any time, because it takes place on the storage area network. Various data replication methods can be implemented, and recovery point objectives and recovery time objectives will determine the technique that needs to be deployed. The storage pool approach provides the opportunity to standardise on tape devices, software, and tape media formats, because the centralised tape library is managed and controlled in a disciplined environment, and the use of the tape devices and media capacity can be better utilised. It also provides a platform on which to recover from system and software failures, human error and disasters, and it cuts down on training requirements and human resources.

The MERIT TS team will take a fresh look at a company's environment and identify areas of possible improvement and all findings would be documented. We would identify business and technology problems and work with the customer's IT department to solve existing problems and come up with a server consolidation solution that will meet the current and future business requirements and budget. We will help the customer to source the right products and oversee the implementation of a solution. MERIT TS can also source, implement and maintain a total solution if required. The implementation would be done in the form of projects and sub-projects to ensure minimal disruption of the business. We will define the strategic scope of the project, draw up a user requirement specification and come up with a high-level design of the server consolidation solution. Our team will carry out an audit of existing storage and backup processes and help implement improvements to ensure the continued integrity of data. Conversion from the old media to the new media can also be accommodated. As product independent consultants, we will assist with the procurement process, which would include analysing possible solutions and carrying out product comparisons on the customer's behalf, and we will assist with the tender documentation and adjudication. Our team has gathered a vast store of knowledge, expertise and experience from implementing this type of solution in many different complex enterprise environments and our methodology has a proven track record. The MERIT TS value proposition is low risk, because we define the deliverables of each project and sub-project beforehand in a documented form and we undertake to transfer the required skills and expertise to the customer's staff. We have had exposure to all the major storage network and backup product suppliers such as IBM, EMC, NetApp, HDS, HP, Adic, and Brocade, but at the same time we remain product neutral. We design and oversee the implementation of server consolidation solutions for both the NT and Unix environments.