Last night I attended and also presented at another excellent SDForum Cloud Computing SIG (Special Interest Group). The topic for the evening’s discussion was Cloud Computing Performance.
There were about 80+ attendees and the discussions, together with Q&A after each presentation were excellent. Many thanks, as usual, to Bernard Golden of Hyperstratus for organizing the event. And thanks also to VMWare for hosting at their beautiful facility in Palo Alto.
BTW, if you haven’t sponsored a SDForum SIG, it’s worth doing, not just because it’s not that costly, but it also helps give back to an organization that really “gets the word out” and provides good old fashion in person networking opportunities.
I presented on behalf of RainStor (disclosure: my current company), together with 2 other excellent companies Riverbed (RVBD) a $360M public company focusing on WAN optimization and Gear6 a Memcache specialist. if you attended and would like a copy of my presentation, please feel free to email me at ramon.chen@rainstor.com
As I had previously blogged about in previous conferences, even if you did not attend the meeting, you could have gotten some interesting nuggets through Tweetlites (my coined term for attending a conference by watching a hashtag for tweeted insights from attendees).
Below are some Tweetlites: For a full list see http://twitter.com/#search?q=cloudsig. Tweets are primarily from myself and @johnmark who diligently tweeted some excellent highlights.
johnmark: biz value in removing hw + sw costs, more immediacy than archival storage, and simplified mgmt tools – @rainstor @ramonchen #cloudsig about 14 hours ago from Identica
johnmark: via @ramonchen – showing some charts showing comparable performance to oracle db and ec2 – is that real data? #cloudsig about 14 hours ago from Identica
johnmark: via @ramonchen – rainstor uses storage compression algorithms – storing only “truly unique values” and their node links #cloudsig about 14 hours ago from Identica
johnmark: from @ramonchen regulatory drivers from govt. pushing data to be produced “instantaneously” or within a few seconds #cloudsig about 14 hours ago from Identica
johnmark: from @ramonchen – rainstor is “making the big data problem smaller” – high perf. storage and query of massive data volumes #cloudsig about 14 hours ago from Identica
Tweetlites About Gear6
johnmark: gear6: does “dynamic” caching, persistence, “full memcached protocol compliance”, web mgmt iface, + analytics #cloudsig about 14 hours ago from Identica
johnmark: gear6 guy: memcached can't handle cloud services – “twitter is a victim of this problem… took them > 2 days rebuild cache” #cloudsig about 14 hours ago from Identica
johnmark: gear6 guy: best practices – cache bidirectionally, design (gracefully) fail, use consistent hashing, use 64-bit, use mysql #cloudsig about 14 hours ago from Identica
johnmark: gear6 guy: going into depth on memcached – not a DB. nor persistent data store. nor large object cache. but highly distributed. #cloudsig about 14 hours ago from Identica
RamonChen: #cloudsig #sdforum Memcache=high perf distributed memory object caching system. Speed up dynamic web apps by alleviating database load about 14 hours ago from TweetDeck
RamonChen: #cloudsig #sdforum Gear6 speed of data to the edge=revenue lost. Cache everywhere in app, webserver and database about 14 hours ago from TweetDeck
johnmark: gear6 guy: for every .1 ms in latency, you lose some $$$ #cloudsig about 14 hours ago from Identica
Tweetlites About Riverbed
johnmark: Riverbed: can improve performance and network latency by at least 30-40%, sometimes much more #cloudsig about 15 hours ago from Identica
johnmark: Riverbed – repacks 64KB window to get around TCP bottlenecks #cloudsig about 15 hours ago from Identica
johnmark: Riverbed – basically an appliance in the data center that analyzes traffic patterns for redundancy to remove extraneous bits #cloudsig about 15 hours ago from Identica
RamonChen: #cloudsig #sdforum data dedupe to remove redundancy for data transportation. Doesn’t Cache instead analyze traffic at disk layer. about 15 hours ago from TweetDeck
RamonChen: #sdforum #cloudsig Riverbed WAN optimization presenting right now. Sftwre on each node perform in harmony about 15 hours ago from TweetDeck
johnmark: “there is a hyperbolic decay in throughput as you add latency” – bob from riverbed #cloudsig about 15 hours ago from Identica
On an amusing note, there were 3 Bernards in the same meeting yesterday, Bernard Golden (our host), Bernard Slede (HP) and Bernard Kyle (RainStor), here is a rare pic of “The Three Bernards”.
One thought on “RainStor, Gear6, Riverbed. A Trifecta Of Offerings for Improving Cloud Computing Performance”