The holidays are upon us and traffic is backing up everywhere. I don’t do much of the Christmas shopping at my house so typically I’m not really impacted. This year is different. The Avere office is near one of the largest malls in Pittsburgh and there are lines everywhere I go. Traffic lights are backed up. The parking lot at my favorite burrito shop is packed. Even the drugstore is full of holiday shoppers.
Waiting sucks.
(Beware of segue.)
And waiting is exactly what you’re doing if you’re using a file server with high latency. Check out Jeff Butler’s blog post on this very point from last week.
Latency can kill your company’s productivity and, more importantly, profitability. Amazon found that every 100ms of latency cost them 1% in sales. Google saw a 20% drop in search usage from 500ms additional latency. TABB Group estimates that if a broker’s electronic trading platform is 5 milliseconds behind the competition, it could lose at least 1% of its flow; that’s $4 million in revenues per millisecond.
In a previous blog post, I explained how Avere is providing a high-performance NFS file server that uses, on average, 5x less disks, power, and space than competitive offerings. Avere demonstrated this using SPECsfs2008 results that can be found on the spec.org site. This time let me use these same results to demonstrate our latency advantage.
Like before, let’s compare all the results that achieved greater than 100,000 ops/sec. (Note: SPECsfs2008 reports latency as ORT, or overall response time, in msec.)
Avere FXT 2500 6 Node Cluster: 131,591 ops/sec, 1.38msec ORT
BlueArc Mercury 100 Cluster: 146,076 ops/sec, 3.34msec ORT
Exanet ExaStore 8 Node Cluster: 119,550 ops/sec, 2.07msec ORT
HP BL860c 4 Node Cluster: 134,689 ops/sec, 2.53msec ORT
Huawei Symantec N8500 Clustered NAS: 176,728 ops/sec, 1.67msec ORT
NetApp FAS6080 FCAL Disks: 120,011 ops/sec, 1.95msec ORT
Avere is the clear winner when it comes to latency (i.e. ORT). The average of all the non-Avere latencies above is 2.31msec. This is 67% higher than Avere’s latency of 1.38msec. This difference can have a huge impact on application performance and the productivity and profitability of your business. For example, this can mean completing a large data processing job in an hour on Avere rather than an hour and forty minutes on average with the other solutions. Add this up across all the people and jobs in your company and this can mean getting your product to market months earlier with Avere.
Don’t wait. Check out the product pages on our website for more information.
Now if we could only get some elves to help with our end-of-year shipments.
SPEC® and the benchmark name SPECsfs®2008® are registered trademarks of the Standard Performance Evaluation Corporation. Competitive benchmark results stated above reflect results published on www.spec.org as of Dec 21, 2009. Above we compare all SPECsfs2008_nfs.v3 results that achieved greater than 100k ops/sec throughput. For the latest SPECsfs2008 benchmark results, visit http://www.spec.org/sfs2008.
