direct compression of all database data  and reducing data where it is the most expensive, tier one, active storage. In this article we are going to explore Storwize’s capabilities as it relates to Data Reducing Oracle.

This type of compression appliance is an inline tool that sits between the application and the storage system. Today the inline systems work on network file systems and network mounted databases.  More and more organizations are moving to NFS mounted databases, especially when served from enterprise class NAS environments like NetApp or Onstor, because of the simplified management and comparable  performance relative to traditional block mounted storage systems.

Rule 1: Don’t affect Performance

While storage savings is certainly interesting, the first and very fair question most IT professionals are going to ask is “what is the performance impact?”  In the end it does not matter how good the compression or  how much money you can save on storage if the companies primary database becomes bottlenecked in compression/decompression performance issues.

Compression appliances from companies like Storwize are specifically designed for this task and special care is taken to maintain line speed performance for both reading and writing. As a result compressing data prior to it being processed by the storage area delivers some unique benefits that would typically not be considered.

First assume that compression is at least 50% for data moving through the appliance, although in actuality in Oracle environments you will see as much as an 90% reduction. An immediate benefit is the reduction in the amount of data that is written to and read from the storage array.  All storage systems, NAS especially, have their own memory, CPU and network resources. By reducing the traffic being written to and read from the device as much as 50% you are in effect reducing the impact on those resources by as much as 50%.

One of those resources, memory or cache, is always in short supply.  With active data compression you are in effect doubling the size of both read and write cache. Read performance is improved as a result of being able to store more upcoming reads in cache. Write cache on storage systems is used to send an acknowledgement back to the application server that the read has occurred and that it is OK to continue processing. Being able to store more data in cache during the write process is going to allow the storage system to sustain high write performance across more simultaneous writes from multiple application servers.

The result of these increased efficiencies in many cases leads to an actual increase in performance when using an in band data compression appliance. As a result an IT professional can have confidence that    these solutions will have no impact on performance of their Oracle application.

Rule 2: Lower Storage Capacity Requirements

With the performance questions resolved we can return to the original question. What are the benefits of using an inline compression appliance in an NFS mounted Oracle environment.

First is the obvious compression of the primary data set. As stated earlier this data set tends to be stored on high speed and high cost disk technology. As we have seen from data deduplication technologies for backup and even when storing Oracle information to tape drives it is some of the most compression friendly data sets in the enterprise. Typical compression savings in an Oracle environment is over 75% for the Oracle Software itself, over 80% for the Data, Temp, Undo, Redo files and over 85% for the Index files. This means that a 5TB Oracle environment can be reduced to less than 2TB’s, saving 3TB’s of disk space while improving performance.

This reduction spreads well beyond just the core Oracle Database. Oracle by its very nature creates a lot of redundant data including copies for test and development. The environment has excellent snapshot tools  so multiple snapshots of the Oracle environment are very common. Lastly this data is typically critical to the hourly operations of a company’s data set so it is replicated through out the infrastructure to make sure that it can be recovered quickly in the event of a disaster.

Having this data in a fulltime compressed format pays off in all of the above scenarios. The same 90% compression factor reduces the redundant test and development data while providing the same performance gain. Because snapshots are also compressed more can be taken, more frequently, and they can be retained on disk for a significantly longer period of time. Also, since the data is already compressed when  transferred across the WAN segment, effectively more than doubling the available bandwidth, replication jobs complete more quickly insuring that the remote sites stay in closer sync with the primary.

Seamless integration

Gains in performance and storage efficiencies are quickly wiped out if integration of the in-line compression systems require a complete rearchitecting of the Oracle environment. With the Storwize system as an example there is no change to the Oracle workflow or the storage itself. It simply sits in between the application servers and the storage and its operation is completely transparent to the rest of the process.

Availability

Keeping data available is always a concern especially in an Oracle environment.  Inline compression appliances can be configured for high availability  by simply attaching a box to each of the redundant network segments you already have in place to maintain network availability.

Enhances the current backup process

Because data is already in a compressed format both backup and restore performance can improve significantly Even if a data deduplication appliance like those from Data Domain or Quantum are in place or are going to be used, this solution enhances those systems as well. By given the deduplication appliances less data to  store and process  you can further reduce the capacity requirements of  those appliances.

For DR it is recommend that a unit be in the DR site as well. It makes sense to do so for the same rapid recoverability reasons stated above. If for some reason a unit can not be made available to the DR site then a software reader can be made available to decompress the data.

Best of Both Worlds – Increase Performance, Decrease Capacity

Inline data compressions appliances not only reduce the data foot print on the most expensive tier of storage, but also for secondary and tertiary storage.  And, they do so while increasing runtime performance as well as backup and recovery performance. They are an excellent strategy for Oracle Administrators looking for ways to reduce costs while increasing performance.