Well, the 2016 UK May user group has finished. It has been a lot of work getting here, and a lot of organisation to get our first 2-day event up and running. Jez commented at the start of his talk on how far the user group has come – six years ago it was 12 people in a room with a bad Skype link, and now we over 100 attendees, nearly 30 speakers and two packed days of user talks, IBM talks and sponsor talks. We even had a VP or two along this year to hear what everyone had to say. We were amazed how fast we “sold out” (we’re free to attend!)!
First, a note of thanks, to Claire O’Toole, our group Secretary (and OCFer!) and Ulf Troppens, our IBM development liaison for getting the event together. And also to all the speakers for coming along, talking to us, listening to us and allowing us to challenge them for the future. We brought together speakers and users from across the globe – three continents (and that wasn’t just the IBM speakers!).
We had attendees from across the whole “Spectrum” of sectors, research institutions, VARs, finance and media, and hopefully the attendees were happy with the programme we put together. It was also great to see some of the L1 support guys from the UK – a great opportunity to put a face to a name. This year was the first time we’ve put together a two-day programme and hopefully people learned new things, found it useful and maybe most importantly, that we’ve inspired some people to look at new features of Spectrum Scale, try things out and upgrade from 3.5!
This year we returned to the IBM Client Centre at IBM South Bank, chatting to some of the speakers, it was the first time they’d been to IBM South Bank, and for some their first visit to London – I hope we impressed them with our event!
A bunch of us met up for an informal dinner and drinks the evening before, which was a great way of meeting people, and checking that the IBM speakers flying in had all made it … Claire and I didn’t have to make any contingency plans for the agenda which was nice!
This year we filmed many of the talks, and we’re hoping to post videos from the talks over the next few weeks – we need to cut them down and plan to fix up the slides, in the meantime, many of the slide decks are already posted on the presentations page.
Day one started with myself and Claire opening the meeting and handing over to Doris Conti (Development Director for Spectrum Scale) to give us our keynote talk. Doris mentioned that she wants the people at the user group to each year to say “Wow” about what they have been doing over the past year, there’s certainly some areas where I think that is the case.
Ulf followed on bringing us an overview of the development priorities for 2016 and was followed by Markus giving us a sneak peak of the next GUI release coming this summer. Just before lunch, we finished with the first of our Sponsor talks – Ray from Seagate talked on Spectrum Scale and IoT, some interesting deployments using Spectrum Scale for Windows as an ingest target for CCTV systems.
After lunch, we were met with another sponsor talk. Jez from ArcaStream talked on their Python API and gave a short demo of it in use. Our first user deployment talk, Dave from MilkVFX gave a fascinating overview of their deployment of AFM between their London and Cardiff offices (and their showreel video got a round of applause as well). Next up, a sponsor talk, Mo from NetApp spoke on the E series systems and how they can be deployed for HPC file-systems.
Our second user deployment talk came from Pam at NCAR, talking on their massive deployment changes happening this summer. The final pre-break talk came from IBM with a deep dive look at the Hadoop integration of Spectrum Scale and how the new transparency layer works. Check-out the slides for details on accessing your POSIX file-system data from your Hadoop workloads!
The final set of talks began with a sponsor talk, with Darren from Mellanox giving an overview on how fabric cards could be used in the future to support storage systems. Another two deep dive technical sessions followed, with Robert talking about the Transparent Cloud Tiering product coming later this year, and then Mathias spoke on Problem Determination, the new features in the 4.2.1 release and the new mmhealth command.
We wrapped up the day on time and then many of us headed down to Namco Funscape in County Hall for the evening social event. The bumper cars and bowling proved a massive hit along with the pool and table tennis tables. Ulf and I even checked out the rollercoaster simulator!
Day two started a little earlier – we figured most people would already be in London, and it was great to see pretty much everyone make it for the 9:30 start, even with the evening social ….
We kicked off the morning with parallel break-out sessions, with a well-attended talk on licensing from David, and an equally well attended sponsor talk from Laurence at OCF talking on Scale out NAS. I was surprised at how few people (OK only me!) have faced the PVU to Socket conversion … the take-away is to start the process early!
Our breakouts continued with Indulis talking on metadata (with real actual numbers and how it works), a session to chat with the GUI developers, and Nils leading a series of Spectrum Protect (TSM!) talks – covering Spectrum Protect Space Management (ILM), Archive (LTFS-EE), SOBAR and using Spectrum Scale as a store for Spectrum Protect. Our other breakout speakers included Madhav on AFM and Olaf and Yohann talking on encryption and compressions. There was also a packed session on monitoring, with Doris opening to the floor for open discussion on what we’d like to see and how to integrate.
Lunch saw the IBM Client Centre people reset the rooms back into a single space for our afternoon slots. Doug asked for a few minutes to present the results of day 1’s survey – how much storage we looked after in the room, and how many people do it … We found amazingly that its 2.9PB/person!
Vic from DDN brought us another sponsor talk, covering adventures in AFM, and looked at their Imperial College London deployment and the massive effect link latency can have on AFM throughput. Gaurang followed up talking on OpenStack integration, covering the Cinder driver and also the object on file work and gave a demo of using object on file with the Manilla driver to have file-based access to a file whilst simultaneously having access to the same file via Swift. The session was wrapped up with Madhav speaking on IBM Life Sciences and how Spectrum Scale can be used to underpin the demanding research needs of the life sciences areas.
Time for another break, followed by Michael from Lenovo, a sponsor talk, clarifying a few post Lenovo divesture myths and talking about development of the GSS solution and how they are still working closely with IBM to deliver the product. We closed the talks with Sven talking about IBM research giving us an insight into what is being thought about for the future and also on some major performance developments which have been driven by major contracts that have a benefit to all.
Doris provided us with some closing remarks and thoughts on the day before Claire and myself wrapped up the day. For me, it has been a great but tiring few days. We’ve had a lot of positive feedback so it was well worth it!
Oh, and we event started discussions about finding a bigger venue for 2017. We’re still hoping to bring together a meet the devs later this year and also hope to coordinate another (much!) smaller user group meeting towards the end of the year.
Finally, a note of thanks to the sponsors for the event, without their support we couldn’t have made the evening event happen.