Unified virtual infrastructure

Introduction
Unified Virtual Infrastructure (U.V.I) is a strategy that combines both Server and Desktop Virtualisation to a preferred unified infrastructure.
With consolidation, power savings and with less administration, virtualisation is at the top of every organisations I.T priorities.
So why look at a U.V.I strategy? The reason is simple, most organisations have virtualised or looking to virtualise, their server farms because the benefits are too compelling not too. So now we see Virtual Desktop being the next big virtualisation project because again, consolidation, power savings with thin clients and I.T control the desktop remotely as well as control software licensing by delivering the right software needed by the different departments.
So what is a U.V.I strategy, this is where both virtualised server and desktop come together on a single unified infrastructure platform, which would be the physical server (cores matched to memory and I/O), storage, networking, protection and security. The infrastructure around these technologies needs to be designed accordingly so that organisations can reap the benefits from delivering virtualisation to the end users and hence V.D.I.
Infrastructure
So organisations will have infrastructure in place already for virtualisation but is it scalable and will it perform to the standards required for a happy end user environment.
The answer is maybe in the short term but this may need addressing when it comes to future proofing and making sure the investment made is not ripped out and replaced within a year, which for tight budgets due to the economic downturn is not an option.
So what infrastructure should be used when looking at U.V.I? Firstly planning virtualisation is no easy task because what you decide you need today and the future changes on a month by month basis due to new systems and expandability you didn’t see coming.
• Networking
With any virtualisation project, the network can be a huge bottleneck due to the different traffic the core switches are being hit by. This could be data, voice, file, print and a host of others, so designing your network is critical for virtualisation to work effectively and where possible having dedicated switches for these network services would be an ideal situation but if this isn’t within the budget then look at having dedicated VLAN’s to segregate the network traffic.
Virtualised I/O is now something that has hit the market, which allows both network and storage connectivity to be managed through a single cable as well as dynamically change network properties on the fly.
With the right U.V.I strategy in place, the networking is designed to allow virtualisation not to suffer.

• Server
The Server estate is always the overlooked part of virtualisation due to server market being commoditised by the large vendors. This is a good and bad thing. Good because organisations can save on capital expenditure, but bad because it doesn’t necessarily mean the servers are going to future proof the virtualisation needs and it might cost more from an operational expenditure point of view due to power, upgrades and administrative costs that were not factored in.
Also when designing the server estate it’s important to make sure that core to memory and I/O is checked and double checked as this is where the headache begins when implementing virtualisation, due to the incorrect measures being made.
So when it comes to implementing a U.V.I strategy you can be sure that the server estate is thoroughly checked so that there is no additional expense in operational expenditure.
• Storage
Storage is another factor where U.V.I can help because data growth is a massive headache, and organisations are hit by continuous additional storage costs.
Storage comes in many different flavours with vendors offering many different functions from multi-protocol, virtualisation add-ins for site recovery and server rollback and both block and file.
So with storage how do you choose the right solution, U.V.I will factor in these choices. So first you need to look at performance from the IOPS (input, output per second) needed, and this is achieved by RAID type and drive amounts (the more drives the more sustained IOPS) generically each virtual server will require 50IOPS so this is a good basis to start with for example 30 virtual servers will require 1500IOPS from disk and this is for sustained read/write and disk cache wouldn’t be factored so don’t be fooled into cache making up this allowance as pre-caching shouldn’t be included, when designing your required storage.
Protocol types, this is always difficult because of cost, iSCSI is a cost effective option but can cause bottlenecks due to the nature of iSCSI bandwidth, latency and limitation in what speeds can be achieved. Fibre channel is great for speed and latency but the fibre channel infrastructure is still expensive and unless you are a huge enterprise the cost of the fabric is generically more expensive then the storage so hence most organisations will go with iSCSI.
Tiers of storage, again this is something important that U.V.I can help with as in an ideal world organisations could buy all Tier 1 storage and have no issues at all but this is at a huge expense. So designing a tiered strategy helps lower costs by dedicating tiers to application types.
Storage functionality is an additional benefit as this removes the administrative headache as well as preserving your virtualised environment. Site recovery management, virtualised backup, virtual machine rollback, application integration, load balancing and a whole host of other plug-ins and functions all help and should be considered and checked as part of your U.V.I strategy.
• Backup, Disaster Recovery, Business Continuity
Backup, Disaster Recovery, Business Continuity are extremely important and are part of U.V.I to make sure the virtualised environment is protected.
- Backup allows you to backup the day to day files and applications of an organisations virtual estate at mainly file or image level but as good as this has been for many years, there is still a window of data loss as generically your backup works after hours and finishes before users come in, in the morning. This allows a certain amount of data loss due to the backup not running in the day. It is still important as with all insurance policies the more you have the more you get back.
- Snapshots, offer recovery above and beyond generic backup. Where backup has a certain window of operation, snapshots deal with securing data in the day when systems are active by snapping the volumes on an off-host basis, therefore securing virtualised environments when they are at their most active but at the same time can restore an environment within seconds/minutes if a disaster or corruption occurs, even to a file level with some vendor technologies.
- Disaster Recovery policies can be many things but the generic meaning of DR is to have an offsite copy of the data, whether this be a tape taken off site or replication from one site to another.
With U.V.I you can use many ways of doing this from block level replication (see also, the storage section), file level continuous data protection/recovery (CDP/CDR) or simply taking a tape offsite every night.
Block level and CDP replication are fantastic for both DR and BC because they can be virtualisation/application aware, so in a disaster of a primary site, they offer crash consistency and can bring up a primary system at a DR/BC site within minutes.
• Security
With your U.V.I strategy almost in place it is important to wrap a security layer around your virtualization environment, strong encryption, access control and key management across platforms, applications and devices. With new legislation being introduced in data security the penalties could be financially crippling to an organisation especially those organisations who house customer databases, financial information and any other confidential data.
• Management
With any infrastructure it is important to have the right management from internal organisational policies to manage procedures to the software stack that services the infrastructure from a configuration and administration perspective. U.V.I takes all the management into account as without it the virtual environment can become difficult to maintain.
• Maintenance
Depending on your business it’s important to make sure you have the right type and term for maintenance. By implementing a U.V.I strategy organisations can shave costs on maintenance due to the infrastructure designed being fault tolerant in all core parts. This means that instead of having a 24 hours by 7 days a week and a 2/4 hour response time, organisations can move to a Next-Business-Day model (NBD) to cut costs of the higher end 24/7/4 offers. This is not for everyone and 24/7/4 in large enterprises is a must have but again the maintenance can be something that when looking at U.V.I can be decided.
• Consultancy
This is always one of the parts of an implementation that is the first to be shaved off, because organisations feel that sometimes the vendor is adding consultancy to obtain more margin and in some cases this is true. With a U.V.I strategy it is important when looking at consultancy to see a statement of works (SoW) for all aspects; this not only protects an organisation but also protects the vendor. So it is important to sit down and review all consultancy and also all vendors (if multiple) into a meeting to make sure your consultancy will meet the criteria set and allows project management to be streamline and to the timeframes set.

Main Menu

See Also

Main Menu