Archive for the ‘Technology’ Category

Docker is not Enough

Friday, June 19th, 2015

I headed up to the London PaaS User Group Meetup yesterday evening. There were two speakers on the agenda. First up was Jeff Hobbs, CTO & VP Engineering at ActiveState with a pitch entitled Docker is not Enough (pretty much this deck). The main tenet being that Docker is not enough in itself as it just addresses packaging and execution. You need a PaaS to provide all of those other niceties like load balancing, auto-scaling, monitoring, centralised logging, audit, … My main issue with this pitch was simply that I don’t think anyone has ever claimed that Docker is enough. That’s why there’s a wealth of ecosystem projects surrounding Docker. And why does stackato (ActiveState’s Cloud Foundry based PaaS) use Docker for containerisation? Jeff stated that this was because the ops team would feel more comfortable dealing with this technology on the back end.

Second up was a late breaking change, my colleague Julz Friedman had stepped in to give a re-run of his Building a Docker backend for Garden presentation from the Cloud Foundry Summit. It was perhaps no great surprise to discover that, when you swap in Docker behind the Garden API, you don’t really see any benefits over the existing implementation (and, indeed, there are significant disadvantages for a multi-tenant PaaS such as the current lack of user namespace support in Docker). The one potential benefit that Julz did highlight was an increase in security, given that there are more eyes on the Docker codebase than Garden.

So why do I make the five hour round-trip in to London for a couple of sessions that I could have got off the internet? Was it the free beer and pizza? Well, no, although welcome, I think the train fare would have more than covered those. It is, of course, to meet people and to hear the Q&A where perhaps much of the interesting information is exchanged (and I caught up on some reading on the train!). There was a lively debate on the relative merits of Docker. One point that Jeff and Julz agreed upon was that the use of Docker images was a retrograde step versus the application centric view of PaaS, letting things that should be the responsibility of ops (e.g. patching OS images) become a part of the developer’s domain (Jeff quoted a stat that some 70% of images on Docker Hub were subject to vulnerabilities).

Docker London

Thursday, January 8th, 2015

Anyone following the WASdev site may have noticed that I’ve been doing some work with WebSphere Liberty and Docker recently. I was therefore pleased to have successfully made it off the wait list in time to travel up to Docker London for my first meetup of the year on Tuesday evening.

The meeting was compèred by Ben Firshman from Docker and, after a mad scramble for the limited amount of pizza on offer for 200 people, the evening began with a short intro to SoftLayer who were sponsoring the venue. Andrew Martin from British Gas was the first of the main sessions, talking about Building and Testing Docker Containers as practised on their ‘connected boilers’ project. I’d seen Andrew speak at Container Camp at the same venue last year so I was glad that he’d included some new material, even if he did then have to race through it a bit. He’d probably have been fine just to cover building or testing rather than both.

Next up was Johan Euphrosine (aka proppy) from Google who demoed a few different ways to deploy Docker containers on Google Compute Engine. Hopefully there’ll be a recording of the event as, whether it was the strong French accent, or too much beer and not enough food, it was sometimes hard to keep up.

Last up was Dan Williams who provided an entertaining and enlightening presentation on what containers are really all about. It was just a shame that, in staying for his talk, we missed the last train before the Basingstoke-Winchester engineering works began and then a freight train broke down at Eastleigh. Suffice is to say that, despite a good evening, I would have preferred to get to bed slightly earlier than 1am!

Countdown Coding

Saturday, August 16th, 2014

I went along to the Southampton Code Dojo on Thursday evening. I think it’s safe to say I was amongst the older attendees (most appeared to be undergrads or postgrads at the Uni although there was one guy who was sufficiently young to need his Mum to accompany him!). A pre-event poll had settled on Java as the language (Python had been outlawed as too popular and I guess Java was the lowest common denominator after that). Following pizza and beer there was another poll to select the challenge: the Countdown numbers game, before being numbered off in to groups. There were only three people in mine with one claiming no coding experience (despite being in the Computer Science department!). We spent far too long looking for an intelligent solution before doing the sums and deciding that brute force would suffice. My brain wasn’t entirely in gear (perhaps due to a day spent with the children?) and we didn’t manage to complete our solution before being timed out. We were in good company though with only one out of the six groups completing the exercise. It has reminded me that coding can actually be challenging in its own right (the challenge at work typically coming from legacy code, integration with other products, or simply politics).

Meetup Happy

Saturday, July 19th, 2014

I’ve gone a bit meetup happy in the past two weeks. Last week I headed along to the Pivotal offices in London for the first London Cloud Foundry User Group meetup organised by one-time colleague Duncan Winn. First to speaker was another ex-Hursley employee, Glyn Normington. He gave a fascinating presentation in to the work that he and his colleagues are doing to replace the backend of Cloud Foundry’s Warden container with libcontainer (now split out from Docker). More on this over on Glyn’s blog.

Next up was London based Tammer Saleh, Director of Products at Pivotal Cloud Foundry Services. You can see the recording of this session from the Cloud Foundry Summit where they talk about the different models for stacking server instances. Finally, James Watters (Vice President of Product, Marketing and Ecosystem for Cloud Foundry at Pivotal) talked about the roadmap for Cloud Foundry in 2014 (including what’s out of scope). See James Bayer’s session from the summit for similar information.

The next meetup was my first at Agile South Coast. If nothing else, this gave me an excuse to have a nose at the new(ish) Ordnance Survey offices! I can’t claim to have been welcomed with open arms to the group (no-one even commented on the fact that they hadn’t seen me there before) but that’s fine by me. Most notable to me though was the fact that I was the only one there who wasn’t a scrum master by profession. Have developers lost interest in agile?

As one would expect with this audience, it wasn’t long before the post-it notes were out and we were collaborating on choosing subjects to discuss. My heart sunk when topics such as “should spikes be given points?” were selected but I was glad when the resounding response from the group seemed to be “it doesn’t really matter – whatever works for you”. Oh, and apparently PSM is more through than CSM but the latter gets more CV points! As I’m part way through reading Kanban in Action, the discussion on Scrum vs Agile in a BAU environment was interesting. I may yet make it to another of these meet ups.

The American style pizza and good selection of beer certainly helped make the trip into town worthwhile although I’ll not mistakenly pick up the 7.2% Sierra Nevada Torpedo Extra IPA in future!

Lastly, I returned to Developer South Coast for a session entitled “NoSQL vs SQL… Fight!”. Actually, there wasn’t much of a fight to be had as the speaker (Tony Rogerson) is an SQL Server DBA. He gave a thorough although halting coverage of the theory behind relational and NoSQL databases though which sadly meant he ran out of time before reaching the potentially more interesting topic of NewSQL databases.

AWS Update

Monday, May 5th, 2014

At the end of April I went to the AWS Summit at ExCeL London, partly to keep up with the competition but largely because the attendees are a different crowd to those you get at your average IBM conference. I managed to miss most of the keynote, partly by design (no early start and an off-peak ticket) and partly due to someone driving in to a level-crossing in Southampton! Having watched the video subsequently, I don’t think I missed a great deal. The only announcements from Amazon that peaked my interest was the arrival of Amazon WorkSpaces in Ireland and the availability of the Twitter stream in Amazon Kinesis.

As in common with the rest of the day, it was the customer slots that were the most interesting. For example, SwiftKey talked about their use of Hadoop on AWS to crunch Wikipedia in other languages to build a starter set for their language models, through to CloudFront as the CDN for serving the final models up to their customers.

I had an interesting chat over lunch with someone who was actually an IBM customer and then wandered the expo watching demos by some of the likely suspects in the cloud deployment, management and monitoring space (Chef, Splunk, DataDog, …).

After lunch the breakout sessions began with six parallel tracks this year. I went to Deployment Done Right first, covering Elastic Beanstalk, OpsWorks and CloudFormation. The only new news for me was an aside that Beanstalk nows supports Docker. It seems like pretty lame support for containerisation though as you appear to get an EC2 instance per image. The accompanying presentation from Sportpursuit.com was most notable for the long list of open source software in use (Nginx, PHP, Magento, Varnish, Redis, Memcached, Elasticsearch, Jenkins, Capistrano, Capify EC2, Boto, …).

Next up was Dynamic Content Acceleration covering the CloudFront CDN and Route 53 DNS with the aim of knocking a second off your response times. The customer this time was import.io which is an interesting site in its own right, providing the capability to turn websites in to structured data (for free).

For the last session of the day I picked Scaling on AWS for the First 10 Million Users which did not, as you might expect, spend a lot of time on auto-scaling, but covered all aspect of application architecture that would contribute to scaling. The customer was the mobile taxi app firm Hailo who are pursuing a micro-services architecture. They are using containerisation (they didn’t specify which) and are apparently writing their own controller to manage the distribution of those containers across EC2 instances to balance workload.

Synchronised training

Friday, May 2nd, 2014

For the past couple of years, uploading my running training has been a bit of a faff. I like to use a desktop app (SportTracks) so that, whatever the changing fads online, I still have all of my data in one place. (It goes back to 2006 when I first got my Forerunner 305. Some day I may even important my old Polar Training Software data.) The desktop app also has some interesting plugins. One of these I used to push my training to dailymile for comparison with a few friends and colleagues and for the widget on this site. However, I’m also partial to a bit of segment stealing on Strava. Sadly no simple SportTracks plugin for Strava so that was a separate upload.

After a bit of search and experimentation, my new workflow is to upload to the SportTracks website which then syncs seamlessly (in both directions) to the desktop app. Then I’m using the excellent online tool Tapiriik to automatically synchronise the data from SportTracks to Strava.

On the plus side, this means I’m now only downloading once and it doesn’t have to be on the machine with SportTracks installed. Also, without really thinking about it, it means I know have 8 years worth of data synced back to Strava!

On the negative side, Tapiriik doesn’t support dailymile so that will have to go by the wayside, at least for now. That means you’ll see a slightly squished Strava widget in the sidebar of this site. The other major downside is cost. The SportTracks site has an annual subscription of $35. Whilst the site is nice and they are continually adding good features, if it weren’t for the sync with the desktop app I wouldn’t be forking out that money. We’ll see how things are going come renewal time. Automatic sync on Tapiriik also comes at a cost but a mere $2 pa for what is a very slick site and available as open source if you really wanted to host it yourself.

Now I just need to decide whether to add Garmin Uploader to my shopping list so that I don’t even need to turn on the Mac! (Works with my Nexus 7 but not my S2 unfortunately.)

Logsearch & Decker

Friday, April 4th, 2014

Yesterday evening I headed up to the London PaaS User Group meeting as there were two Cloud Foundry related sessions on the agenda. First up was David Laing talking about the open source Logsearch project, a bosh deploy of an Elastic search ELK log analysis cluster. His employer (City Index) has this hooked up to Cloud Foundry system logs and, in some cases, they’re also using it for analysis of application logs with addition parsers. They’re looking for people to get involved in the project and help with the next phase: anomaly detection. One major hole in the solution as it currently stands: it’s only suitable for private PaaS as their is no access control over the logged data.

Up second was an entertaining pitch by Colin Humphreys, Founder and CEO of ours hosts CloudCredo, on how to sell hats to monkeys. That was the back story anyway, it was actually about how there is space in the stack for something that gives you the flexibility of IaaS over what you run but the simplicity of management, scaling and load balancing of PaaS. That something is Container as a Service. Specifically, the ability to push Docker files to Cloud Foundry using a custom stack for the DEA. Something that Colin is referring to as Decker.

Colin gave a nice demo but it is obviously still early days. Currently you can only push Docker files not images. There is also no staging at the moment – the image is created when each instance starts – consequently it is not taking any advantage of intermediate images. There is obviously lots of scope for improvement and it’s definitely one to watch. It was also interesting that Colin is currently focussing on the Docker side with the DEA interactions set to change with the introduction of Diego. The project is open source but Colin recommended waiting until he writes some docs before you try picking it up!

WebSphere Appliance Management Center

Tuesday, April 24th, 2012

I haven’t posted anything product related since the release of WebSphere ESB 7.5 in June last year where I hinted at pastures new. Those of you that I’m connected to on LinkedIn will know that I subsequently became Technical Lead for a product called WebSphere Appliance Management Center. WebSphere Appliance Management Center provides off-box management and monitoring of multiple WebSphere DataPower SOA appliances. The development for the management component of the product moved to Hursley shortly after the initial release (named 4.0 after the coincident firmware release). The new team put together a fix pack later in the year which delivered support for the new Service Gateway XG45 appliance.

Today saw the announcement of the next chapter in the WebSphere Appliance Management Center story. Significantly, you will find this in the WebSphere DataPower Appliances firmware V5.0 announcement letter as the next version will no longer be a chargeable product in its own right. Instead, it will be freely downloadable and supported free of charge for all customers with a current support entitlement for a WebSphere DataPower SOA appliance.

Not only will it cost you less – but we’ll also be giving you less! The team has been working flat out to create a much lighterweight offering which is faster to install and less resource intensive. The management user interface has also been extensively reworked to be more responsive and better support user interaction patterns. In particular, the current restrictions around managed set membership will be lifted allowing much great flexibility for firmware and configuration deployment. In addition to the existing domain management capabilities, you will also be able to manage configuration at the service level. All of this will be available for download on June 26.

For those attending IBM Impact next week who’d like to find out more about this exciting new release, I’ll be co-presenting on WebSphere Appliance Management Center (Monday, 5:15-6:30pm, Lando 4305). We will also be hosting a series of round table sessions in Toscana 3701 (Monday, 10:45-12:00; Tuesday, 1:20-2:45pm; Thursday 3:15-4:30pm) which will be your chance to shape the future direction of the product. Alternatively, please feel free to contact me directly or post questions in the DataPower developerWorks forum.