Fixing Storage Performance In Virtualized Environments
In our last entry we identified just a few of the reasons why server virtualization projects cause storage performance problems. In the next few entries we will discuss ways to fix storage performance in virtualized environments. In general you have two options; get faster or get smarter. Depending on your environment one option will typically make sense over the other but eventually as the virtualizatio
In our last entry we identified just a few of the reasons why server virtualization projects cause storage performance problems. In the next few entries we will discuss ways to fix storage performance in virtualized environments. In general you have two options; get faster or get smarter. Depending on your environment one option will typically make sense over the other but eventually as the virtualization project scales you will likely need your storage infrastructure to be both smart and fast.The get faster option involves making the whole storage infrastructure fast enough so that it can keep up with any performance peak that the virtual server infrastructure can throw at it. As we will discuss in our upcoming webinar "Stopping The Storage Roadblock To Server Virtualization", if the storage infrastructure can provide more performance than the virtual environment needs, it may also be the simplest option. You are not typically going to fine tune something when no one is complaining.
There are plenty of fast storage networking options available to IT managers today. The problem is of course being able to afford them. If it is a new technology there is also the cost of replacing what you have and learning something new. However if you are at the point of a storage refresh or if you are willing to consider a new storage architecture to support the server virtualization project then the get faster options is certainly worthy of consideration.
Getting faster means you are only as slow as your weakest link. For many data centers that means speeding up the connection from the server host to the storage. In Fibre Channel (FC) environments this can mean upgrading to 8GB/s fibre or even 10Gb FC over Ethernet (FCoE). For NAS or iSCSI environments this typically means upgrading from multiple 1Gb Ethernet connections to 10Gb Ethernet. While the FC upgrade options may be the most expensive they do deliver the purest bandwidth. If a convergence project is also underway then FCoE can be justified for other reasons beyond just the performance boost. NAS and iSCSI still bring the overhead of IP. While IP overhead is not necessarily impossible to overcome, it needs to be factored into the actual performance gain.
There are cost effective options for performance gain that still deliver pure bandwidth improvements. One potential get faster solution to consider is second generation of Serial Attached SCSI, 6Gb/s SAS often called SAS-2. As we point out in our white paper "The SAS-2 Primer" the basic networking and automated use of multiple 6Gb/s channels available on SAS-2 host bus adapters may make it an ideal storage infrastructure for server virtualization in small to mid-range data centers. Another option is ATA over Ethernet (AoE). This technology transports native storage traffic across standard Ethernet cables, there is no IP to SCSI conversion required. Like SAS-2 it can increase bandwidth by using multiple ports simultaneously.
Making the storage infrastructure faster is a straightforward way of dealing with the performance challenges caused by server virtualization. In addition, new architectures have reduced the cost to implement a higher speed infrastructure. However, if you have just purchased your storage system or are not in a position from a budget perspective to replace it then you need to get smarter, something that we will talk about in our next entry.
2014 Next-Gen WAN SurveyWhile 68% say demand for WAN bandwidth will increase, just 15% are in the process of bringing new services or more capacity online now. For 26%, cost is the problem. Enter vendors from Aryaka to Cisco to Pertino, all looking to use cloud to transform how IT delivers wide-area connectivity.
The UC Infrastructure TrapWorries about subpar networks tanking unified communications programs could be valid: Thirty-one percent of respondents have rolled capabilities out to less than 10% of users vs. 21% delivering UC to 76% or more. Is low uptake a result of strained infrastructures delivering poor performance?
Join us for a roundup of the top stories on InformationWeek.com for the week of December 14, 2014. Be here for the show and for the incredible Friday Afternoon Conversation that runs beside the program.