High Availability Storage Options and Their Impact on Performance

To further reduce the need to go to disk, MPE employs a cache as part of its design. This design uses

available main memory to keep active parts of the file accessible by an approach that is similar to direct

memory access. As active applications require access to their data those requests are satisfied through

MPE's memory cache and are handled by the MPE/iX memory manager facility.

The next technique employed by MPE/iX to increase performance is that of reducing the need to post data

to disk. MPE's journaling file system is called the Transaction Manager (XM for short). This facility

copies just those bytes that change in a file, because of a file record update, and copies them to XM's

facility. (This XM protection is afforded to critical MPE/iX structures and some file types like KSAM and

Image data bases.) This reduces the need to post files/data to disk to ensure data durability and this

posting action can significantly impact the overall system and application performance.

The last couple of techniques I want to mention is the way MPE stripes data files across multiple disk

members within a volume set and how it sends multiple I/O request for any given Ldev (volume) at the

same time. This helps reduce the impact of disk seek and latency times, part of the normal performance

impact when dealing with disk drives. MPE bundles a number of I/O requests (8 per Ldev) for most types

disks and arrays and sends them asynchronously. This lets the operating system return to other tasks while

those operations complete.

How is performance perceived

The number one question I usually get is "how fast will this HP e3000 server run with that storage

array?". The answer is, "It depends"! Without any performance data or an understanding of the current (or

future) application's I/O characteristics, that question really can't be answered except very conservatively.

Using a car story analogy; A customer goes to a car dealer looking for a car that has a 400 horse power

engine and "price isn't an option" says the customer. The dealer sells the customer a real nice sports car

but is dismayed when two days later the car is returned dirty, smoking and badly wrecked. The customer

wants his money back because, it seems, the car was unable to navigate off-road, down hill and with a

large boat attached (which explains the dents on the rear and roof of the car). <Enough of cars now>

Without understanding what the needs are and understanding how the technology works together, we are

all destined (in our small way) to repeat this story.

Understanding the Technology

The benchmark for storage technology is set by the performance of JBOD (Just a Bunch Of Disks).

Remember too, that modern storage arrays are built on top of JBOD but use a memory (cache) and a small

operating system to add features that ultimately allow for the added protection of data and in some cases

improve ease of management and performance.

The added value that arrays provide over that of JBOD is that of data protection. It does this by laying the

data across a number of disk drives along with data recovery information that can be used to reconstruct

the data in the event that a drive mechanism fails. The vender may have created a way for the

reconstruction to happen on the fly or as a background activity. The main point to remember is that

whatever technology is in place, the goal is to protect the data.