ERDAS Image Web Server Blog

Why compress?

You may have noticed that over the past few years the sizes of hard-disk drives has grown tremendously.  If I look on Amazon.com today, i can see 1.5 Terabyte hard-drives for around $150.  Wow!  That is pretty cheap.

So, you may ask yourself, do I still need to compress imagery?  I have got these massive hard-disks, maybe I don't have to worry about it.  But there is something you need to be aware of....


Disk speed is pretty much the same as it was 10 or more years ago

Overlooked by many, particularly hard-ware vendors and IT departments that like spending  (your) money, is that whilst there has been massive growth in the capacity of hard disks drive, they are pretty much the same SPEED that they were years ago.  Speed increases have been a lot slower in coming about than the increase in disk size. 

There have been some "trivial" movements with "SCSI" and SATA drives.  5400 or 7200 revolutions per minute compared to 10,000 or 15,000.  Not much of a comparison when you compare it to 80GB to 1.5 Terabytes.

The constant rate of disk speed is no better emphasized that with "seek time".  "Seek time" is the time it takes to "find" your information on your disk.  This has stayed "static" for the most part in the last 10 or more years. 

The relative slowness in hard-disk speed  means that, among other things it will still take the around the same amount of time to "read" your images today as it did 10 years ago.  

This means that the larger the "data" on the disc, the slower it will be to "read", regardless of having a "big hard-disk".


How does compression help?

So where does this leave "compression"?  Well, there are a few factors that help speed up imagery access when using compression

1. Less data needs to be read from disk....

>> Read more

Asynchronicity for Scalability!

WMS or Web Mapping Service is a popular way of distributing geospatial data. 

However, something that has been showing up more recently is that people may overlook the performance gains they get with simply using client-side asychronous access.  Instead, they make one of their server components work really hard and subsequently slow down their entire application.

Cascading Services through a single server.





In the diagram you can see one WMS Server requesting information from another WMS Server, then sending the combined information onto the client application.  Whilst this may seem "convenient" to a system administrator, the solution is also likely to underperform.  One service is going to be dependent on the other service, and do some unnecessary extra work.  On top of this, often the requesting server doesn't just pass through the information, it saves the information to its own system, then passes on the data.  This doubles up the work needed to be done to present the information. 

A better way to get the data to people faster, is to make "asynchronous" requests to the different services, from the client.

Asynchronous delivery of geospatial data



In this instance, neither service is dependent upon each other, and neither service has to do more work than it needs to.  By saving your system resources, you get to save time, with faster responses.  And, by conserving your system resources, you can do more with your existing hardware....

>> Read more