OneFS is a marvel – and the Isilon team should be proud. It’s delighting customer after customer – including anyone who uses an Xbox and Xbox Live to stream content:
Or check out this story from the Vatican (which not only uses Isilon, but also uses the ViPR controller and Object stack, and Atmos to preserve the contents of the Vatican Archive):
When it comes to Scale-Out NAS – Isilon has 3x greater customers than the next closest competitor, and is growing faster. When it’s about non-scale out NAS and transactional NAS, VNX has surpassed the next closest competitor and is growing faster.
It’s a great time to be an EMC NAS customer! As always – there’s lots to do to continue to improve (Isilon customers, see recent IB switch ETA, VNX customers – see the VNXe post), but overwhelmingly a positive overall state of the union.
The Isilon team isn’t standing still, either. You can expect the following in the future:
- Faster, Bigger. Some colossally big.
- More and more about combined NAS/HDFS use cases – this is proving to be very popular. In almost any enterprise – their dataset doesn’t all make sense in HDFS, or all in NAS.
- More and more about Sync n’ Share integration with Syncplicity (already there – which is really, really cool!)
- Architectural improvements in the kernel, abstracting filesystem services and more.
- NDU – forever. Even customers who LOVE Isilon still wish they could count on NDU (it’s not a bad situation today, but could be better).
- Software-only vOneFS (some of Isilon’s biggest web-scale customers are asking for this) – this is more than the VSA that is available now (to ALL here)… It’s really looking at the IO stack and determining how to do it best (NVRAM dependency – can we do this another way? Should we look at other things like data distribution?)
- Embedded OneFS – Scale-Out NAS appearing in different form factors.
- CloudPools.
This last one is really cool, and will appear first as a preview feature this year. We demoed it at EMC World this week – take a look below:
Are you an Isilon customer? How’s it going? (give me the good/bad/ugly!)
If creating a single file pool across multiple sites, how does performance become affected, I know when we break this down it will be about latency. But lets for a moment consider that File A is in the Cluster furthest away from me, once I access and retrieve this data, will the oneFS be smart enough to move File A to the cluster closest to me so access next time is not constricted by latency?
Posted by: Storageous.wordpress.com | May 07, 2014 at 05:43 AM
If you place workload in the cloud in essence you're less bothered by latency... which is why a lot of webapp tend to be developed with cloud framework. Does it matter when you watch a video online is the film is in a remote location? No because you buffer it for rapid local access. If data must be local, set isilon policy for the file to remain local to the user or application. Alternatively use caching policies to store most recent access data locally.
Posted by: Fe | May 09, 2014 at 06:05 AM
We are really looking forward for Multisite OneFS that would be super cool. On the other hand we are very curious if also the small things get fixed and improved, like the missing "NFS sub-directory visible" setting in SnapshotIQ or Zone features for NFS to support different NFSv4 NAmespaces.
Posted by: Andy | May 15, 2014 at 11:12 AM