High Availability Architectures (4/4) - Technology Trends
In the previous episodes of this serie, we’ve addressed the Availability, Scalability and Performances aspects of HA Architecture. In this one we’ll concentrate on the future of these architectures and the emerging technologies to tackle specific HA constraints.
Technology Trends
The future is a G word : GRID. Grid of memory, grid of CPU and Grid of disks. The main limitations today is to have this one to one relationship between the application and the physical server on which the former is deployed. Hence the main trend of the market today : to virtualize servers by using Network Access processing, memory and storage.
Network Accessed processing
This is the Grid of CPUs. Azul Systems offers some sort of Java mainframe, a box containing 768 CPUs and 700GO of memory. Applications are deployed on blades as usual but these blades contain a proxy to the Azul box : whenever CPU process is required, the blade proxy hands it over to the Azul box which is configured to allocate a certain number of CPUs for that very app.
Websphere XD also offers new possibility on CPU and servers virtualization.
Network Accessed Memory
Terracotta offers a solution for Network Accessed Memory. This is a server managing objects lying in network memory. Thus differents applications running on different JVMs and different servers can share the same instance of a given object. Client applications just need to import the terracotta client libs and define in a description XML file the objects and attributes to be shared and that’s it !
Main issue here with the open source version : 2mn start up time. This would then create a main Single Point Of Failure in the system.
Network Access Storage
SAN (Storage Access Network) offers a very robust and efficient solution for network repository. Communication are fiber channel based and therefore very performant .. but very expensive. This already is commonly used and it has paved the way for the above 2 other solutions of network access services.
I have a question about your choice of vendors. So are these the top-of-the-line manufacturers of hardware for NAP, NAM, and NAS respectively? Or the ones you are familiar with?
How would one go about implementing technology-solutions? Would you suggest going for an integrated solution of hardware + software, or buy them separately and build software on top?
Comment by Vincent van Wylick — November 9, 2007 @ 2:56 pm
Hi thanks for the mention about Terracotta. I think a correction is in order because there is not a 2 minute start up time, and there are no SPOFs in the architecture.
Please feel free to email me for more details - I would like to know how you came to this conclusion.
Full disclosure - I work for Terracotta.
Comment by Taylor Gautier — November 9, 2007 @ 4:59 pm
You see Taylor, this is exactly the sort of comment that makes me love blogging. The web conveys such a proximity between people! It’s just amazing for you two guys to interact and share opinion and expertise. My take is that Cecil felt like giving just an overview rather than detailed specs on Terracotta, so I guess, since you work for your company, that you’re right.
Anyways, I’ve printed out, read and underlined all 4 articles of Cecil on high availability architectures: a really good job at making complex technology accesible to all!
Comment by Jeremy Fain — November 9, 2007 @ 9:39 pm
Vincent,
For NAP the actual concept has been “created” by Azul System for their system doing an analogy with NAS. So I guess that for the time being they’re the only one for sure the first ones.
For NAS/SAN there are quite a few solutions around I am not an expert on the field.
For the NAM field I guess Terracotta have probably one of the most mature solution on the market. GigaSpace also offers a solution but our consultant was not as enthusiast as he is for Terracotta. We actually had a demo with the Terracotta product : 2 instances of a Java Gui running in 2 different JVMs : a draw on one GUI would appear instantly on the other one. VERY impressive.
As far as this solution is concerned and to answer Taylor (thanks for your comment Taylor) maybe I should have said that the consulmtant reported us that the main limitation was this 2 minutes start-up time. I didn’t try myself.
By the way Taylor, if you have this NAM solution on one server taking care of instances of distributed objects, dont you identify this as a SPOF ?
Comment by ceciiil — November 12, 2007 @ 2:35 pm
[...] de Facebook comme monstre tentaculaire et omniscient, comme matrix doucereuse : dans le cadre de billets techniques sur des architectures informatiques à haute disponibilité (le sujet super interessant, déjà) [...]
Pingback by Back on the HTML gang « Heavy Mental — November 13, 2007 @ 10:58 pm
ceciiil,
The Terracotta server itself can be clustered, which today can be configured as an active passive system. In the event of a failure of the active server the cluster will automatically switch over to the passive server with no loss of data or functionality.
In addition to this configuration, Terracotta supports persisting all data to disk which provides even more ability to intentionally or unintentionally terminate any process or all processes in the system with no loss of data.
Hope that helps - we can always move the discussion somewhere else or forums - I don’t want to hijack your article entirely
Comment by Taylor Gautier — November 14, 2007 @ 6:55 am
Sorry to post one more time - your consultant is correct there is a configurable timeout that is 2 minutes by default but only happens to protect against a double failure - it’s an edge case that makes sense when explained in detail - I’d be happy to explain but fear this is not the appropriate venue.
Comment by Taylor Gautier — November 14, 2007 @ 7:01 am
Taylor, many thanks for your feedback. Jeremy is right regarding the proximity between people blogging conveys.
I have to say I was quite impressed with the Terracotta demo. I didn’t know data could be persisted. But again the aim here was not to write a comprehensive description of that solution but rather to introduce the different solutions for NAM, with special focus on Terracotta which is seen as a leader on the market.
Thanks again for your comments they are fully appropriate, dont worry about hijacking the article.
Comment by ceciiil — November 14, 2007 @ 11:01 am
Thanks for the interesting article Ceciiil. I work for Azul and it’s great to see people blogging about our technology.
I also wanted to bring one clarification on our solution. Our compute appliances deliver both CPU and memory capacity as a ‘Network accessed service’. In many ways we are the ‘power station’ for the grid. Any Java-based application hosted on a conventional server can access this capacity transparently over the network.
It’s also interesting to note that we work closely with the vendors you identify as ‘network accessed memory’ (such as Terracotta). As a matter of fact we just announced a strategic alliance with one of these vendors two weeks ago. You can look at Azul as the provider of raw CPU and memory capacity, while these software vendors provide the data management layer that leverages this capacity to provide a full ‘data grid solution’(data scalability across multiple nodes, data access, resiliency, HA, replication).
Comment by Gaetan Castelein — November 27, 2007 @ 4:00 am
Excellent article, Ceciiil. Thank you.
Comment by Pavan K — January 21, 2008 @ 1:59 am