The first day at the Next Generation Storage Summit, focused on Object Storage, is just finished and there are a few interesting things that I noted.
1) Object storage vendors are now targeting large enterprise customers, not only verticals or large ISPs. This is a first step towards maturation of the entire segment but, as I said, this is only the first step. The technology will be ready for the prime time only when we will have solutions for a wider range of end users.
2) APIs suck! This is not news actually but, eventually, everyone here agrees on the fact that the average enterprise can’t, or don’t want, to cope with APIs. They want something very similar to what they are deploying today and network file sharing interfaces (SMB/NFS) are in pole position. This is another step for a broader adoption.
3) End-user cases have demonstrated that object storage could be interesting for medium sized end users too. We should talk more about this.
4) There is a wrong misconception (on the vendors side too) about how to measure the performance of an Object Storage system. They often use IOPS instead of something like Object Operations Per Second (OOPS?) and throughput, for example. This opens a big discussion on how to make benchmarks on Object storage (but this is a long story).
5) There is a certain general acceptance about Amazon S3 imposing the de facto standard for APIs. All vendors are working on it. They also provide proprietary APIs and CDMI… but the demand for S3 compatibility is always higher.
6) The number and the kinds of gateways available on the market is quickly increasing, giving the end users plenty of choices to build their next generation infrastructure.
Why it is important for you
Object storage, as an industry, is slowly maturing.
I think object storage should be a fundamental building block in any serious private cloud infrastructure, especially the bigger ones.
Technology is maturing while, at the same time, user needs are increasing. This two trends are set to cross each other very soon.
Disclaimer: I was invited to this meeting by TruthinIT and they paid for travel and accommodation, I have not been compensated for my time and am not obliged to blog. Furthermore, the content is not reviewed, approved or published by any other person than the Juku team.
Love the info!
Thanks!
what’s difference betwwen IOPS and OOPS?
First of all, OOPS is not an official term I used it only to describe the concept.
From my POV IOPS are useful when you are talking about blocks but not in this case. when you perform an object operation (like put, get or delete for example), this operation can be extremely amplified in the backend and objects could be very different (think about the difference between a movie and an email).
I was only trying to say that we should find a better way to benchmark OS, during the event Intel talked about it but I missed the presentation. :/
See, thank you for your reply! Will Intel and other speaker put there presentation or slides on line?
You can find the videos here http://truebittv.truthinit.com/media-gallery-media.html
There is also a page with the presos but I do not have the link now