Database Systems of the Future

Database technology has remained somewhat unchanged for a long time, especially when it comes to data protection and data recovery. The near-future of database technology, however, will bring many welcomed changes to help reduce the Recovery Time Objective (RTO) and Recovery Point Objective (RPO) all database technicians struggle with on a daily basis.

Traditionally database teams create database copies and mirror them to achieve the RTO and RPO they need. Technicians typically make a point-in-time copy and then back it up to disk or tape. They employ mirroring-techniques to increase these copies (perhaps in multiple locations) to protect their databases from catastrophic failures. These systems are generally referred to as "storage-centric" recovery schemes. A de-facto standard for many years, storage-centric recovery and protection schemes are simple but limited. They are not aware of the underlying data structures and consistency points of the database, as they blindly copy at a low level. This ultimately results in tedious manual intervention after a failure, for steps must be taken to perform crash-consistent recovery before the database can go online again, thus lengthening recovery time.

Storage-centric backups do not recognize physical or logical failures either. Corrupted data could potentially be mirrored into backups without the backup system even recognizing it, compromise recovery and lengthening the before a database can be put online again. They are also severely bandwidth inefficient, as they cause a lot more write operations than are necessary because they do not understand the database's underlying data structures. Consumption of more bandwidth reduces available bandwidth for "real" tasks and increases telecommunication costs as well.

Telecommunication costs aren't the only expense to cause database technicians grief. The cost of replications, snapshots, and every other operation involved in Continuous Data Protection (CDP) is expensive due to storage vendors charging big bucks for the system as a whole. The situation worsens for the technician as vendors almost always play the "hardware lock-in" card on their customers. Most turnkey storage-centric solutions require customers to use all components from the same vendor and product family. This puts a strain on a database team's budget, as they are not free to shop around for the best deals. Ideally technicians should be able to choose high-end components for their primary storage solution and cheaper components for secondary and tertiary ones. With vendor lock-in this simply isn't possible.

Integrating data protection (DP) and disaster recovery (DR) within the database itself can help resolve many of these problems. Database-integrated disaster recovery and data protection techniques are data aware, meaning they understand the the data within a database and how it is structured rather than backing it up blindly. This provides more optimizations that storage-centric solutions cannot address. Created snapshots do not require crash recovery procedures. This technology allows a DBA to, at any time, rollback modifications anywhere from the table level all the way up to the entire database. It is also possible to provide immediate disaster recovery to a live standby database, eliminating lengthy recovery procedures.

Data aware DP/DR systems are hardware agnostic, meaning they can work seamlessly with any "mix-or-match" storage system. They are very bandwidth efficient, too, only propagating select changes in an incremental fashion, thus saving bandwidth and money. There are no distance limits either, as replication techniques residing within the database itself are not limited by Fibre Channel limitations. They are free to send data worldwide using WAP IP links.

Such data aware systems are a database technician's dream, solving problems that have been almost career-long problems for database teams worldwide.

Copyright 2010 All Rights Reserved