Your virtual world resides in a physical world, so it is considered to be as safe; however, in this Q&A English explains that not everything from a physical world can reside in a virtual one. Here are some considerations you need to take in to account when choosing hardware for your virtual infrastructure.
What should an IT manager consider when deciding on virtualised server hardware?
Ed English: It's wise to choose a vendor that has a broad choice of servers, across blades, classic racks or a combination of both, to avoid being locked into one strategy.
Previously, it was the bigger businesses that would use centralised storage; however, that threshold is coming down.
Ed English, product manager at Dell,
You need to choose the right type of server for you, with lots of I/O flexibility and ideally a server that has room for PCI cards, too. Look for those that have Ethernet and infiniband option to avoid any lock-in.
Memory footprint is also an important deciding factor. How many virtual machines do you want to put on your server? This is normally affected by the amount of physical memory you have, so it's something to bear in mind before buying or you may find yourself having to buy separate memory later down the line.
You also need to consider how you want to apply a hypervisor. This normally happens on hard drives, but they run a lot power and are known to fail. Many hypervisors are now embedded in SSD cards/disks, as these are proving more reliable and more cost effective.
How much of an impact will my choice in storage have on my virtual infrastructure?
English: Before you buy, decide how you want to deploy your storage. Many businesses are now opting for centralised storage instead of deploying it directly on to each server. Here the image is not stored at the server level and is therefore not dependent on whether or not the server works, making this method more robust. This also means that you have all your data sitting in centralised storage, so any server can access it at anytime -- it doesn't have to be the one attached.
Previously, it was the bigger businesses that would use centralised storage; however, that threshold is coming down. Now even if a business has 50 staff members, a data backup and recovery process still has to take place at night, and this takes longer with direct storage. So centralised storage is starting to look more and more attractive to all sized businesses. Centralised storage offers ROI in just months, not years, with the amount of time it can save IT staff on backups alone.
What would be on your check-list for organisations looking to deploy or upgrade servers to specifically support virtual infrastructures?
- Choice of server: Blade, classic rack or a combination of both.
- Storage: Should you look into something more advanced than direct storage for your business needs?
- Networking infrastructure: Are you considering migrating to 10G Ethernet soon? Would you prefer Fibre Channel or are you leaning more towards Ethernet?
- Management: Less physical servers tends to mean more virtual servers. This means new virtual tools for management, new ways of managing software licenses and new ideas needed to ensure you don't get locked in to a vendor that has a very complex infrastructure.
All of these things should be simple to manage, not complex.
What type of organisation would want to choose blades?
English: If you only have about 10 servers, then it's probably best to stick with rack servers. However, if you have more servers, then blade can have some great benefits. If you are space constrained, for example, then blade is a good choice for you because of the densities you can get with blade systems.
If you want to consolidate your switch infrastructure, too, then blade is a good choice for you. Other systems can look messy, with lots of cables, but with blades there are less. This also reduces the amount of power needed for cooling, as the data centre is clearer.
With blades it is easier to implement new functionalities across businesses that have several locations. For example, if a supermarket across Europe decided to implement a new scanning functionality where the customer scanned items themselves as they shopped and were charged for all the items as they left the store via their credit card, this would only need to be implemented centrally. Instead of having to individually visit all the stores across Europe, it can be configured centrally and deployed to all of the supermarket's locations.
Kayleigh Bateman is the Site Editor of SearchVirtualDataCentre.co.uk.