“Caché offers all the characteristics sought by those in the NoSQL movement.
CACHÉ 2010 Interview
JAXenter asked InterSystems' Robert Nagle a few questions about the recent CACHÉ 2010 release, and its new CACHÉ Database Mirroring functionality and CACHÉ eXTreme for Java solution.......
JAXenter: The CACHÉ 2010 release introduces CACHÉ Database Mirroring. What is this new feature?
Robert Nagle: Database Mirroring is a new alternative for high availability (HA). It offers equivalent levels of HA to existing configurations but at lower cost and with less complexity. Database mirroring can also be used to offer disaster recovery (DR.)
Database Mirroring uses redundant servers and redundant storage in place of shared storage and complex cluster management software. With the plummeting cost of servers and continued decreases in storage costs, this architecture can be significantly less expensive.
Logical replication between the servers relies only on a TCP connection between them. Mirroring offers "warm" failover and failback.
A summary of the benefits can be seen at the Intersystems website (PDF.)
JAXenter: How does the new CACHÉ eXTreme for Java solution, CACHÉ bring and Java together?
Robert Nagle: Caché has long offered developers a powerful development model for complex data. Persistent data can be manipulated as objects (the most natural paradigm for developers) or as SQL tables/views (the lingua franca of reporting). A frugal kernel has consistently delivered extraordinary performance and extreme scalability. This means that less hardware is required to deploy the Caché based solution. Furthermore, Caché deployments are simple to manage and don't require constant care and feeding by DBA's and system administrators.
Until recently, to garner these development and deployment benefits, developers had to use Caché's server side script language. With Caché 2010, we've now made these benefits available to Java developers.
With Caché eXTreme for Java, Java developers can tunnel directly into the Caché kernel to manipulate complex, persistent data as objects or SQL.
JAXenter: What particular benefits can CACHÉ eXTreme offer Complex Event Processing applications?
Robert Nagle: Caché eXTreme is particularly suited for accommodating very low latency, very high volume incoming data rates. Caché offers the speed of an in-memory database but as a "real" database, the objects/tables are persistent. This avoids the resiliency/reliability issues associated with in-memory solutions.
JAXenter: In your opinion, what benefits can object database such as CACHÉ, offer that relational databases cannot?
Robert Nagle: In complex data modeling, it is difficult and unwieldy to map the "natural" form of the data structures onto relational databases. For such applications, an object model is far more natural.
Of course, all persistent data immediately becomes the target for reports and queries so that it is vital to have high speed SQL access to the data. Caché offers both rich, natural persistent object storage and blindingly fast SQL access to the same underlying data.
JAXenter: Who is InterSystems targeting, with CACHÉ 2010?
Robert Nagle: Many of the capabilities in Caché 2010 are targeted at (and often developed in conjunction with) our existing customers. For example, database mirroring will be a boon to many of our largest ISVs and direct end users.
InterSystems is also, with the introduction of Caché eXTreme for Java, targeting Java developers who are tackling problems for which they either have to develop an in-memory storage framework today or have to select a highly customized solution to deal with low latency, high volume data acquisition problems.
JAXenter: NoSQL is a hot topic in the database world at the moment. What is your opinion on the NoSQL movement?
Robert Nagle: NoSQL is a movement that started in response to many of the perceived disadvantages of relational solutions. Relational technologies are very powerful for certain classes of problems. However when development or deployment models call for characteristics such as lightweight or flexible or high levels data complexity or minimal administration, relational solutions are rarely appropriate.
For three decades, InterSystems has been focused on providing technologies to developers with exactly these characteristics. Since its introduction in 1997, Caché has offered precisely these behaviors.
I believe, however, that NoSQL is a misnomer. I find that persistent data immediately becomes the target for queries and reporting. And the most widely used technology for queries is SQL. The unsuitability of relational technologies for certain operations does NOT imply that we should reject SQL. As proof of this, we are starting to see some "NoSQL" solutions beginning to add SQL query capability.
Caché offers all the characteristics sought by those in the NoSQL movement. AND it offers blindingly fast SQL.