Topics:
Storage
Understanding Storage Bandwidth Performance
For most data centers we can get more than enough storage bandwidth performance. There is 8GB fibre, 10GB FCoE and 10GB Ethernet. The first step should be to make sure that the application can take advantage of that performance if you upgrade to it, if not then don't. This is one of the reasons why smaller environments are served just fine by iSCSI even on 1GB connections, for them that bandwidth is good enough. Understanding the speed at which I/O is coming out of the server to the storage connection is key. Again as mentioned in our last entry, some of the built-in utilities with the OSs can give you this information. As the environment gets more complicated, server virtualization for example, then we suggest using tools that can give you a more accurate determination like those from Akorri, Virtual Instruments or Tek-Tools. In addition to the application, you also need to examine the server hardware itself. Does it have the CPU power and memory size needed to get the data requests to and from the storage network? In most cases the chances are yes. If you have an application that needs high I/O you have already upgraded the server hardware. In the case of virtualized server environments, can that storage bandwidth be channeled or better allocated? Server virtualization changes the rules. Instead of a single server accessing storage or the network through a single interface, now you have multiple servers all accessing storage simultaneously. As we discussed in our article "Why Quality of Service is Even More Important in a Virtual Environment", optimizing high bandwidth in virtual environments may be better served when priorities can be given to each virtual machine. When it comes to I/O at the storage system most systems today have multiple connections leading up to the switch infrastructure, or in the case of clustered storage the connections to the switch infrastructure scale as more storage is added. It is important there is plenty of bandwidth to the storage controller since it is likely to receive storage requests from many servers and virtual machines simultaneously. Most storage managers overbuy on storage system bandwidth and it is typically not the primary issue in the performance loop. The disadvantage to this is that you are often paying for bandwidth that you will never use or at least not until a future date. Ideally, you should be buying just the bandwidth you need today and then upgrading it when you need it. This is one of the deliverables of clustered storage systems like those from 3PAR, Isilon and HP's Lefthand Network’s because bandwidth can grow as you need it, providing maximum flexibility and CAPEX optimization. Clustered storage also plays a big role in addressing the next area of performance concern, the storage controller itself, which we will discuss in our next entry. Track us on Twitter: http://twitter.com/storageswiss Subscribe to our RSS feed. George Crump is lead analyst of Storage Switzerland, an IT analyst firm focused on the storage and virtualization segments. Find Storage Switzerland's disclosure statement here. « Services To Keep Data Flowing On 4G Networks | Main | Vivisimo Search Engine Aims For Scalability » |
| Sign Up Now For InformationWeek News Alerts |
This is a public forum. United Business Media and its affiliates are not responsible for and do not control what is posted herein. United Business Media makes no warranties or guarantees concerning any advice dispensed by its staff members or readers.
Community standards in this comment area do not permit hate language, excessive profanity, or other patently offensive language. Please be aware that all information posted to this comment area becomes the property of United Business Media LLC and may be edited and republished in print or electronic format as outlined in United Business Media's Terms of Service.
Important Note: This comment area is NOT intended for commercial messages or solicitations of business.
