XIO Virtual Business Intelligence Storage Analysis

One of the most reliable of those surveys, conducted by The Data Warehousing Institute (TDWI) identified a number of drivers that are leading organizations to consider replacing their existing primary data warehousing platform.

Interestingly, five of the top six responses -- poor query response, lack of support for advanced analytics, inadequate data load speeds, inability to scale data volumes, and exorbitant scaling expense -- are all related, fairly directly, to the capability of the underlying storage infrastructure.

Storage, it turns out, is the number one limiting factor in effective data warehousing and business intelligence initiatives today.

What is the meaning of Big Data? Data is now arriving continually -- streaming, rather than being fork lifted into the data warehouse each night via extraction, transformation and load (ETL) tools.

There are far more active, demanding users connected to the warehouse than there were even a few years ago -- and more and more of those users are programmatic, not human. Operational decision-making is increasingly being handled in code to ensure that those decisions are made in a timely fashion: increasingly, within seconds of a state change in the data warehouse.

The sustained query rates that warehouses are expected to support -- while they're loading data in near real time into very complex schema -- is higher than it ever has been.

All of these trends -- more data, more frequent updates, more users, more complex schema and higher sustained query rates -- exhaust the IOPS in the data warehouse's storage later sooner.

In the last 20 years, disk capacity has increased three thousand times and disk performance has only doubled. This imbalance between servers and storage is only going to increase as processors continue to get faster and add more and more cores.

There are a number of ways to increase performance but they all come at a cost. Enterprise customers are literally spending millions of dollars to solve DW/BI performance issues.

Choosing to abandon the commodity DBMS and SMP platforms your organization understands and can manage effectively, in favor of an exotic MPP architecture with a few dozen users and a spotty track record of success, is a high-risk way to get the IOPS your data warehouse needs.

Hyper ISE provides extreme perform with price/performance that changes the economics of the data center. Hyper ISE can achieve up to 200,000 IOPS in a single 3U storage system with 14.4 TB of capacity. Additionally, XIO customers can aggregate multiple Hyper ISE systems using our software, X-Volume provides near linear scalability. Therefore, the performance of Hyper ISE continues to increase meeting the needs of any high performance environment.

Hyper ISE is the only True SHD storage system that fuses SSD and HDD as single pool of storage creating the ideal balance of price/performance and capacity.

Hyper ISE can achieve up to 200,00 IOPS with a 3U and 14.4 TB of capacity. A single Hyper ISE matches the sweet spot for Oracle and SQLServer data warehouses today: in the 2-10 TB range. And, for larger warehouses Hyper ISE can scale linearly into a multi-petabyte pool of storage for the largest warehouses in the world.

Hyper ISE allows you to do what you already know how to do: make commodity DBMSs and SMP servers perform better, for large DW environments. There's no need to take on the time, cost and risk of exotic MPP environments to get the benefits of higher aggregate IOPS: Hyper ISE gives you those benefits in the environment you manage today.

The Hyper ISE storage system was designed for performance. And here is something that you must take into account -- most vendors provide you performance information based on "best-case-scenario". However, their performance degrades as they reach high levels of capacity utilization; during RAID rebuilds, performing data movement, mirroring or when they have any drive issues. However Hyper ISE has been architected to minimize any performance degradation that is felt by the users or applications during any or all of these conditions occurring. Hyper ISE is fast forever.