Case Studies
Eliminating Database Overload at an Online Testing App in Publishing
In the use case described by the above image, the typical user has a workflow that exists across many HTTP requests to a load-balanced application cluster. That workflow needs to persist in case of server failover. The initial approach was to store workflow in the database. The 3 silos below show how Terracotta replaces the database for storing the in-process workflows and provides workflow persistence without any extra code. The application performance increased, and the user in this case, avoided a million dollar clustered database upgrade.
Avoiding Network Overload at an Internet Portal in Online Media
In the use case described by the above image, the typical user has a cache that he wants distributed to many JVMs / hardware machines on which the application has been installed. The value of distributing the cache is in the simplicity of being able to just place objects in cache and not worry about if those objects are in synch with other JVMs. The system will also run faster for end users if the cache can stripe across servers such that 5 servers, each with 1GB heap can in fact store 5GB of data striped instead of 1GB of data, duplicated. The initial approach was to store the cache on its own server cluster and then scale using replication with a solution like JGroups or JMS. The 3 silos below show how Terracotta replaces the custom clustering solution and caching subcluster, thus striping gigabytes of cache across a few low-cost hardware machines. The application performance increased by 100X, and the user in this case avoided months and months of hand tuning and application stability issues.
Terracotta Distributed Cache Performance Case Study
This case study presents the results of an evaluation of Terracotta vs. a leading commercial distributed cache product, performed by a Fortune 100 Company. In the key areas of performance, simplicity and cost, Terracotta outperformed the alternative by a significant margin. The Fortune 100 Company expects to realize significant performance improvement and cost savings by incorporating Terracotta into their architecture.
Read the case study >>
Need for High Availability at an Online Gaming Site
A large web community that allows users to build up a network of friends where they socialize and compete in an online reality game played for fame and fortune was running on a single server. With already nearly one million users, the community has received wide acclaim for its innovative model that combines social networking with game dynamics and gives people within the social network a way to gain status by winning contests.
Their rapid growth forced them to seek a seamless scale-out solution, but they couldn't afford to invest time developing it at the expense of new feature roll-outs. Terracotta's true POJO clustering and distribution transparency was a perfect fit to deliver a solid scalability and H/A story in a very short timeframe.
Highly Scalable Back-End for Massive Web-Based Games and Applications
SmartFoxServer is a comprehensive and powerful platform for rapidly developing multiuser applications and games with Macromedia Flash MX, MX 2004, 8, Flex 2 and Flash CS3. The integration of SmartFoxServer with Terracotta opens unlimited possibilities for creating the next-generation of highly scalable back-ends for massive web-based games and applications.
The great flexibility provided by this combination allows anyone, from small startups to larger enterprises, to benefit from the clustering services at a fraction of the cost of traditional solutions, and with best overall performance.
Additionally the non-intrusive nature of Terracotta allows the developers to concentrate on their application logic, without the need of introducing and learning new dependencies.
Read the whitepaper for an in-depth look at building highly available SmartFoxServer clusters with no single point of failure, high horizontal scalability and failure resilience.
High Availability and Scalability for iPoint Portal
iPoint portal is an open source, standards compliant portal server developed by C2B2, experts in SOA, JEE and Portal Solutions. C2B2 used Terracotta to cluster iPoint, allowing iPoint to be deployed in a high availability configuration and scaled on demand by effortlessly adding new iPoint server instances.
Read how a cluster of iPoint portal instances sharing a single database was built, using iPoint 2.4.2, Java 6.0, Apache Tomcat, Hibernate, Terracotta and MySQL to provide a scalable architecture for iPoint deployments.
Fast, Reliable Asyncronous Database Replication for Large North American Lottery System
In this case study, the customer requires replication of database writes from a primary to a secondary database for high availability. This replication can be asynchronous, but must be reliable and adhere to a specific SLA. Using MySQL database replication, while reliable, was too slow for the use case. The steady state of 700 tps without MySQL replication plummeted to 100 tps with MySQL replication enabled.
The customer used Terracotta to fork the writes at the application level so that each write to the database was accompanied by a write to a Terracotta durable shared queue. A second JVM was used to take writes from the Terracotta queue and perform that write job against the secondary database.
In the event of failure of the primary database, the application can automatically fail over to the secondary database. To ensure data integrity in a failover scenario, any pending writes in the Terracotta write queue are consulted prior to the secondary database until the write queue is drained.
The writes to the Terracotta queue are ACID, therefore meeting the replication reliability requirement. Because the writes are asynchronous, the customer was able to achieve the SLA of 700 tps with replicated writes to the secondary database.
|
