Service Technology Symposium 2012 – Talks update 1

Building Cloudy Services;

The main premise of this talk was that you need to understand the cloud paradigm when designing services that you plan to run in the cloud.  Everything you do in the cloud costs, minimise unnecessary actions and transactions.

Why is the cloud an attractive solution? – Cloud computing characteristics..

–          Uses shared and dynamic infrastructure

–          Elastic and scalable  (horizontally NOT vertically!)

–          On demand as a service (self-service)

–          Meters consumption

–          Available across common networks

Features you should consider for any services that will be hosted in the cloud; where + indicates patterns / beneficial designs, indicates ‘anti-patterns / designs that will be more challenging to run successfully in the cloud;

Failure aware;

–          Motivation – Hardware will fail, software will fail, people will make mistakes

–          + stateless services, redundancy, idempotent operations

–          stateful services, single points of failure

Event driven;

–          Motivation – No busy waiting, less synchronisation

–          + everything is a call-back, autonomous services

–          – anti patterns – chatty synced interactions, guaranteed latency

Parallelizable;

–          Motivation – horizontally scalable

–          + stateless, workload decomposition, REST/WOA (not SOAP)

–          SPOB (single point of bottleneck), synchronous interactions

Automated;

–          Motivation – Easily recreated and installed (automated provisioning and scaling)

–          + re-startable, template driven

–          complex dependencies, hardware affinity

Consumption awareness;

–          Motivation- Efficient resource usage

–          + fine grained modular design, multi-tenancy

–          Monolithic design, single occupancy

The talk concluded with the following recommendations;

–          Stop treating cloud like a generic shipping container – be cloud aware

–          Match your goals for cloud computing to essential characteristics

–          Promote patterns among development team (paralysation etc)

–          Hunt down anti-patterns in code reviews

–          Evaluate IaaS and PaaS providers based on their support for cloud aware patterns

–          Balance the patterns

Keeping these points in mind should help ensure the services and designs you migrate to the cloud have a better chance of success.

PDF of the presentation can be found here;

http://www.servicetechsymposium.com/dl/presentations/building_cloudy_services.pdf

———————————-

High Performance Computing in the Cloud

This talk was one of my favourites, and somethign I find very interesting.  Traditionally High Performance Computing (HPC) has been the preserve of large corporations, reasearch depatements or governements.  This is due to the size and complexity of computing environemtns required in order to perform HPC.  With the advent of HPC in the cloud access to this level of compute resource is becoming much more widespread.  Both the cost of entry and expertise requried to set up this type of environment are lowering dramatically.  Cloud service providers are setting up both tradtional CPU based HPC offerings, and the newer, potentially vastly more powerful, GPU (Graphics Processor) based HPC offerings.

Onto the talk;

Cloud HPC can bring HPC levels of computational power to normal businesses for things like month / year level processing, and risk calculating etc.

In order to think about how you can use HPC, look to nature for inspiration – longest chain – how small a pieces can a process be broken down into in order to parallelise it?

–          Traditional HPC – message passing (MPI), head node and multiple compute nodes, backed by shared storage.  Scale issues – storage performance (use expensive bits)

–          Newer HPC, more ‘Hadoop’ type model, data stored on compute (worker) nodes – they then just send back their results to the master node(s).

  • Look at things like hive and pig that sit atop hadoop. More difficult to set up than MPI.

–          Newest HPC – GPU – simpler cores, but many of them.

  • CPU – ~10 cores maximum.  CPU – hundreds cores (maybe thousands). 
  • Some super computers looking at 1000’s GPUs in a single computer.
  • 4.5 teraflop graphics card < $2000!!

Cloud scale vs. on premise –

–          On premise = measured by rack at a time. 

–          Cloud = lorry trailers added by simply plugging in network, cooling and power then turning on, left until enough bits fail, then returned to manufacturer..

Cloud = Focused effort! – Cloud power managed by CSP, researchers work.. No need for huge amount of local infrastructure.

How to move to the cloud, largely as with other stuff –

–          Go all in – pure cloud. –

  • MPI cluster – just have images of head and compute nodes – scale out.  10 node cluster hosted on amazon made top 500 computer list with minimal effort in setup / config.
  • Platform as a service – e.g. Apache Hadoop – based services for windows azure – just go through how big you want the cluster through the web interface – has excel interface already so excel can directly use this cluster for complex calculations!

–          Go hybrid – add compute nodes from the cloud to existing HPC solution (consider – latency issues, and security issues (e.g. VPN to the cloud)).

You really don’t care about how the technology works.  Only how it helps you work!

Dan Rosanova who gave the talk has an excellent Blog post with some metrics around HPC in the loud here;  http://Danrosanova.wordpress.com/hpc-in-the-cloud

The slides from this talk can be downloaded here;

http://www.servicetechsymposium.com/dl/presentations/high_performance_computing_in_the_cloud.pdf

Final note – GPU development is currently mostly proprietary and platform specific. Microsoft is pushing their proposed open standard that treats CPU and CPU as ‘accelerators’ it does abstraction at run time rather than compile time.  This would allow much greater standardisation of HPC development as it abstracts the code from the underlying processing architecture.

These are exciting times in the HPC world and I’d expect to see a lot more people / companies / research groups making use of this type of computing in the near future!

K

Service Technology Symposium Day 1..

So yesterday was day one of the Service Technology Symposium.  This is a two day event covering various topics relating to cloud adoption, cloud architecture, SOA (Service Orientated Architecture) and big data.  As mentioned in my last post my focus has mostly been on the cloud and architecture related talks.

I’ll use this post to provide a high level overview of the day and talks I attended, further posts will dive more deeply into some of the topics covered.

The day started well with three interesting keynotes.

The first was from Gartner covering the impact of moving to the cloud and using SOA on architecture / design.  The main points of this talk were understanding the need to move to a decoupled architecture to get the most from any move to the cloud.  This was illustrated via the Any to Any to Any architecture paradigm where this is;

Any Device – Any Service – Any Data

Gartner identified a ‘nexus of forces’ driving this need to decouple system component;

–          Mobile – 24/7, personal, context aware, real time, consumer style

–          Social – Activity streams, Personal intelligence, group sourcing, group acting

–          Information – variety, velocity, volume, complexity

–          Cloud services

In order to achieve this, the following assumptions must be true; All components independent and autonomous, they can live anywhere (on premise or in cloud), applications must be decoupled from services and data.

They also highlighted the need for a deep understanding of the SOA principles.

The second keynote speech was from the European Space Agency on their journey from legacy applications and development practices to SOA this was titled ‘Vision to reality; SOA in space’.

They highlighted 4 drivers for their journey; Federation – Interoperability – Alignment to changing business needs / requirements (agility) – Reduce time and cost.

And identified realising these drivers using SOA, and standards as outlined below;

Federation – SOA, Standards

Interoperability – SOA, Standards

Alignment to business needs – SOA, Top Down and Bottom up

Reduce costs – Reuse; SOA, Incremental development

Overall this was an interesting talk and highlighted a real world success story for SOA in a very complex environment.

The third keynote was from NASA Earth Science Data Systems.  This provided an overview of their use of SOA, the cloud and semantic web technologies to aid their handling of ‘big data’ and complex calculations.  They have ended up with a globally diverse hybrid cloud solution.

As a result of their journey to their current architecture they found various things worthy of highlighting as considerations for anyone looking to move to the cloud;

–          Understand the long term costs of cloud storage (cloud more expensive for their needs and data volumes)

–          Computational performance needed for science – understand your computational needs and how they will be met

–          Data movement to and within the cloud – Data ingest, data distribution – how will your data get to and from the cloud and move within the cloud?

–          Process migration – moving processes geographically closer to the data

–          Consider hybrid cloud infrastructures, rather than pure cloud or pure on premises

–          Security –  always a consideration, they have worked with Amazon GovCloud to meet their requirements

To aid their move to SOA and the cloud, NASA created various working groups – such as – Data Stewardship, Interoperability, semantic technologies, standards, processes etc.

This has been successful for them so far, and currently NASA Earth Sciences make wide use of SOA, Semantic technologies and the cloud (esp. for big data).

The day then moved to 7 separate track of talks which turned out for me to be somewhat of a mixed bag.

Talk 1 was titled ‘Introducing the cloud computing design patterns catalogue’.  This is a relatively new project to create re-usable deign patterns for moving applications and systems to the cloud.  The project can be found here;

www.cloudpatterns.org

Unfortunately the intended speaker did not arrive so the talk was just a high level run through the site.  The project does look interesting and I’d recommend you take a look if you are involved in creating cloud based architectures.

The second talk was supposed to be ‘A cloud on-boarding strategy’ however the speaker did not turn up, and the organisers had no idea if he was coming or not so wasted a lot of peoples time.  While it’s outside of the organisers control if someone arrives or not, they should have been aware the speaker had not registered and let us know rather than the 45 minutes of is he, isn’t he, we just have no idea that ensued..

The third talk was supposed to be ‘developing successful business plans for cloud computing projects’.  This was again cancelled due to the speaker not arriving.

Talk 2 (talks numbered by my attendance) was a Gartner talk titled ‘Building Cloudy Services’.  This was an interesting talk that I’ll cover in more depth in a following post.

Talks three to five were also all interesting and will be covered in some more depth in their own posts.  They had the below titles;

Talk 3 was titled ‘HPC in the cloud’

Talk 4 was titled ‘Your security guy knows nothing’

Talk 5 was titled ‘Moving applications to the cloud’

The final talk of the day was titled ‘Integration, are you ready?’  This was however a somewhat misleading title.  This talk was from a cloud ESB vendor and was basically just an advertisement for their product and how great it was for integration. not generally about integration.  Not what you expect from a paid for event.  I’ll not mention their name other than to say they seem to have been inspired by a piece of peer to peer software.. Disappointing.

Overall, despite some organisational hiccups and a lack of vetting of at least one vendors presentation, day one was informative and interesting.  Look out for more detailed follow up posts over the next few days.

K