Real-life experience of private cloud technology

There are many many articles written extolling the benefits of ‘cloud technology’. However, most of these are written by a supplier or specialised consultant – meaning they aren’t independent. There are also reports written on the subject by consulting companies, but these tend to be theoretical and ‘up in the clouds’ (if you will excuse the pun) – they aren’t based on real-life experience.

What is a ‘private cloud’ ? There are many definitions, but here I am referring to the use of virtualised processing and storage technology for your companys use. This can be hosted on your site, or hosted with a service provider for your own use. Many of the points mentioned are also relevant to public cloud services.

Here are some real-life benefits of using the technology and some advice.

To anticipate one question up-front, I am not affiliated in any way to any product or service provider. My (day) job is to manage IT resources and services for a very large financial services company. My comments here are based on my real-life experiences.

The benefits of using private cloud technology

Firstly, there are the obvious benefits that you can easily visualise. For example ….

• A ‘private cloud’ allows you to gain the benefits of the technology without the security concerns you may have with public cloud services.
• Reduced space in your racks. Note that this may not necessarily mean that your rack floor-space reduces – that depends on whether you can virtualise enough physical server equipment to remove an entire rack.

If you host rack space externally and pay per rack, you can potentially reduce your costs if you can free-up an entire rack of equipment (don’t forget that some suppliers offer half-rack solutions)..

• Less physical equipment = less power consumption, lower costs and good for the environment.
• Less physical equipment = less heat generated, so the air conditioning units use less power consumption, hence lower costs and good for the environment.
• Less physical equipment = reduced maintenance costs.
• Backups can be centralised and taken from the centralised storage instead of the individual servers.
• On the assumption that your solution includes storage replication, application availability is improved because recovery time from an outage is reduced.

But the other benefits aren’t so obvious ….

• Traditionally, providing disaster recovery/contingency/replicated systems that can be used in the event of an incident are a pain ! They need specialist replication or mirroring software, duplicate hardware and constant monitoring. With a replicated SAN, that headache disappears, because the data at your primary site is automatically replicated to your contingency site.

Not only is this built-in replication an operational relief, but it means that when you deploy new services, there is far less infrastructure work to do, because the server and its data is automatically replicated.

• With physical servers, you may have several functions running on a server, to save on the physical servers you needed. With virtual servers, you can create a virtual server for each function without the previous overhead. This makes upgrades and maintenance on the applications easier.
• You will read from product vendors that efficiency benefits mean your System Administrators can manage up-to three times more systems because they are virtualised. Be careful about using any headcount savings when justifying your investment. This is because it depends on the size of your IT team, how many physical servers will be virtualised and what else the System Administrators do.

Remember that these benefits are largely around the physical activities. In each virtual server ‘container’ is the logical server that was previously running on its own physical equipment. By ‘logical server’, I am referring to the business application; for example it could be Microsoft Windows 2008 server with SQL Server and a business application. This still needs as much support as when it was a separate physical instance.

• The most significant benefit I have seen is more important than reducing costs. It is the fact that your IT Infrastructure team’s focus becomes less about engineering and more business focused. This is because there is less physical infrastructure to support and maintain.

Real-life advice and lessons learnt

• Get the design right up-front. This technology will be the core of your business’ infrastructure – if you skimp at this stage, you will regret it later on and it will be very difficult to correct it.
• Work with experts. The skills needed for this technology are very different to previous technologies. Don’t assume your team can ‘pick-up’ the skills ‘as they go’.
• If you don’t have a preference for the product set, try and use a supplier with no bias. This will ensure you get the right product, rather than one that they sell.
• Meet the technical consultants who will actually be doing the work before signing with a supplier. Interview them to understand their skills and previous site references. It is really important that you can work with them on such an important project.
• Make sure the skilled technical consultants work alongside your team to ensure knowledge transfer.
• Send your staff on the necessary technical training courses to be able to support the hardware and software. These courses can be expensive, so beware and budget for them as part of the project.  Don’t go live before this – doing so is a risk to the business.
• Use proven software and hardware. It will be very tempting to use the latest version of something, but remember that this technology will be at the heart of your infrastructure, so make sure it is stable.
• Make sure you have contractual support for all the elements, such as hardware and software support. This is one system where you are likely to have 365*24 support.
• Don’t be  persuaded to virtualise everything. There is strong technical opinion that the primary domain controller (PDC), Citrix servers and SQL Servers should not be virtualised – of course everyone has their own opinion on this !.
• Go live with low-risk systems first, such as test systems. If you make a mistake, it won’t be public and you can resolve any issues in a less-rushed and less-stressed manner.
• Order more disk space than you think you need. You will soon use it with unplanned backups,  snapshots and data that ‘slipped through the net’ when you did your sizing estimate.
• Configure and test the contingency aspects of your installation before ‘going live’. If you discover that you need to make significant changes (for whatever reason), it is much easier to make them on a system that isn’t yet live, than on a live system when you need to plan it around the availability for business use.
• As with all projects, don’t be tempted to jump-in and start straight away. Regardless of special deals, supplier end-of–month special deals and so on, it is far more important to make sure you are comfortable with the supplier’s terms and conditions. Once you get started, you have far less room for negotiation.

Also read ….

Cloud Computing – what is everyone so excited ?