May 2017 UK User Group Report

OK, OK, its been a while and I apologise for the delay in getting this written up (sorry Claire!), so many things have got in the way in my real job!

So the UK meeting this year saw a move from London to Manchester, but still saw a strong following of Spectrum Scale UGers attending. First off, lets talk about the elephant, or rather pillar in the room…. yes the conference venue had THE largest concrete pillars in the space … something to do with it once being a 1970’s office block so I’m told. But we managed, and DDN’s Vic even had some fun with laser pointers and the mirrors mounted on the pillars.

The day kicked off with myself and Mike Daubman from IBM opening the sessions followed by, now regular, Ulf Troppens speaking on Scale updates over the past year and coming up soon.

We then moved on to a couple of sponsor talks – IBM covering the new ESS products (there’s new spinning and flash based systems due RSN) and Rosemary from Ellexus giving us a fascinating insight on how their software tools can be used to identify rogue jobs and provide some level of QoS type support at the file-system access layer, for example alerting when a job is running over x MB/sec of bandwidth.

Thanks then go to Kieren from EPCC for his talk on the RDF with its 23PB of capacity… (but noting they are still on 3.5 code!).

Vic Cornell (DDN), always an audience favourite, then gave us his views on the joys of AFM and an overview of a joint DDN and IBM project running AFM and object on a global scale!

Carl Ziete then followed up with a surprising fun talk on licensing. Myself and Ulf debated including on the agenda, and Ulf was pleasantly surprised at how well what could be a tedious topic could go down. Perhaps Carl’s most important take home is that data management edition is licensed per TiB of NSD capacity. So TCT, tape layers are NOT per TiB licensed. ONLY storage that appears as NSD devices.

Jez from ArcaStream then followed up with a talk and demo of their metadata capture and search tools, originally designed for media, but of potential value to research organisations.

Day 1 wrapped up with two more IBM talks, one from Simon Lorenz on the mmhealth and problem determination work which has taken the forefront in development, and Olaf Weiser with his best practice talk (he even talked us through an example on how a node “UK” could rejoin a cluster known as “EU”)!

Once the talks had wrapped, we then headed over to the Museum of Science and Industry, for networking, drinks and dinner. We were lucky to be housed in the revolution gallery including a replica of the World’s first commercial programmable computer – Baby. The evening event is courtesy of our fantastic sponsors, without whom we wouldn’t be able to run such a great event!

Day two opened with our break out sessions with three parallel tracks of talks covering a range of topics from data protection to cloud integration. I only wish I could have attended more of the talks than I had time for!

After lunch we reconvened into the main space (somehow the pillar shuffled along the space … or maybe the chairs?). Our afternoon session included two sponsor talks, one from Mellanox on using RDMA over converged Ethernet and another from Seagate and their development work on an IO manager to do some auto-tiering type functions for storage.

We also had two user talks, a fascinating insight from Mark Roberts at AWE on his use of AFM to move data between sites and deploy new storage systems, including all the minor and silent issues they encountered, and perhaps the best talk of the day (OK so I’m biased 😉 ), from me on our cloud deployments and scale out storage systems.

I also can’t emphasise how important it is for users to volunteer to speak at our events, a number of people commented on how they’d like more user talks, but this of course relies on getting people to actually volunteer to speak. And people aren’t just interested in the mega-scale deployments, but also small stuff, how its being used and the challenges people are finding.

We wrapped up with the ever popular Sven with his research deck, I’ve never known Sven get more than 50% of the way through his charts in his slot 🙂

Thanks again to all our sponsors and to IBM for supporting the event and to Claire O’Toole (OCF) and Ulf Troppens (IBM) for helping me organise and make the event a success.

We’ve already booked a London venue for 18-19th April 2018, so see you then!