AppLogic vs. EC2: a more reliable platform
I came across an interesting blog article today about making a reliable MySQL database cluster on EC2. The author goes over the pain of working around two limitations that EC2 has: lack of immutable (unchangable) storage associated with an EC2 instance, and the lack of a static IP address for an EC2 instance. However, EC2 isn’t the only game out there for utility computing. Services such as ENKI's Computing Utility, based on 3Tera's AppLogic grid operating system, can provide scalable utility-billed cloud computing at higher levels of reliability than EC2.
With AppLogic, you get 3-nines “right out of the box”, and 4 or 5-nines can be achieved using some of the techniques mentioned in the blog I was reading. AppLogic’s 3 nines comes from its ability to automatic restart on a different, working server any virtual servers that were hosted on failed hardware. AppLogic is oriented around “Applications” which are networks of Appliances in a private address space - essentially a virtual private data center. Even if an Appliance is restarted after a failure, it gets the same DNS name inside the Application that it had before (as maintained by AppLogic) so other Appliances in the Application can continue to refer to it successfully. The automatic switchover usually takes just a few minutes after a server fails.
Where AppLogic goes beyond EC2 is that virtual server images, called “Appliances” in AppLogic, are object-oriented in the sense that the configuration information for the server (application, setup files, patches, etc) is maintained separately from the server instance itself, so if the server fails, it’s restarted from its parent “class.” Clustering or parallel instances are supported because each Appliance's unique configuration information (including config file settings entered from the AppLogic user interface) are loaded into the Appliance when it restarts, yet all the Appliances in a cluster can share a single NAS Appliance and hence share a single deployment of code or data. To get better performance in, for example, a Java/Tomcat environment you need only add another Tomcat instance to your Application.
User data is stored in a RAIDed volume or on a NAS appliance with its own RAIDed volumes within the grid of redundant servers that AppLogic runs on, which means that it is available without interruption even if a server goes down. Again, this storage is more reliable (and available at a lower cost) than Amazon's S3 offering - and it's within the same grid of hardware, connected by gigabit links for better performance.
There are currently 3 ways to use AppLogic. and ENKI can help you get started with any of them. You can purchase utility computing from ENKI on a pay-as-you-go basis which allows you to use a virtual private data center (AppLogic Application) and scale it on demand. In this case, you don’t need to know anything about AppLogic to use it since we manage it for you. You can also lease a stack of servers from a managed service provider offering AppLogic and manage it yourself or hire us to manage it for you. Or, if you’re working for a big enough company, 3Tera will lease AppLogic licenses to you for use in your data center, and we can supply training and support.
TrackBack URI for this entry
