VMware's SpringSource grabs a GemStone
Relational in the cloud
SpringSource has bought distributed, in-memory caching specialist GemStone to help deliver what it calls the middleware of the future.
On Thursday, the VMware unit announced it has acquired both the technology and people behind Gemstone, so that its open-source Java middleware can store, query, and serve vast amounts of frequently accessed data used across very large data centers.
GemStone uses a distributed data fabric that's capable of querying data, while providing messaging, complex-event management, partitioning, and replication of data without the choke point or cost of a mainframe or centralized Unix box. GemStone has 200 customers in financial services and banking, with growing uptake in the US government. GemStone is in talks with the Department of Defense's IT arm. Financial terms of the deal were not disclosed.
The deal follows last month's acquisition of RabbitMQ, the messaging system SpringSource bought with Rabbit Technologies. RabbitMQ is based on the free and open AMQP.
Rod Johnson, SpringSource general manager and creator of the Spring Framework, told The Reg: "We view our middleware investments as investments in the future."
GemStone will be integrated with Spring Framework and SpringSource's Apache-based tc Server. Johnson promised tc Server would scale to hundreds of nodes for cloud-scale computing.
"We are trying to weave together middleware assets relevant to the cloud," Johnson said. "It will be possible for us to add GemStone capabilities in our platform as a service and the GemStone capabilities will be an integral element of our private cloud for Java."
SpringSource is not trying to build what he called a "backward looking middleware portfolio copying what the legacy vendors are doing."
The deal appears to pit Spring against Oracle, which has expanded its middleware with - yes, you guessed it - two acquisitions: data-cache came with Tangosol in 2007 and an in-memory database arrived with TimesTen in 2005.
SpringSource's move comes amidst the rise of NoSQL, with various non-relational architectures are trying to solve different parts of the problem of storage and frequently accessed data in the cloud.
The trouble is that generally, NoSQL architectures do not provide query out of the box. But - increasingly - customers are using NoSQL as a storage system. Last month, memcached specialist Gear6 said it was trying reconcile that by adding native query to its Memecached distribution, turning its emecached into what it called a "NoSQL-like store."
While Johnson has promised the middleware of the future, he has clearly recognized that to get there, Spring must drag a heap of the past with it, and that this past is relational.
"The NoSQL space is a very interesting space that's undoubtedly becoming more important, but relational databases are not going to go away," he said. "Most of those [NoSQL architecture] are somewhere you can place data at rest but does not support transactionality.
"GemStone can sit in front of data stores of record and provides transactional semantics. NoSQL doesn't to that." ®
Sponsored: Benefits from the lessons learned in HPC