During peak hours, we see quite a bit of database activity.
Our customers’ vehicles are reporting every minute or two typically, and they are querying to see where they’re at using maps, reports, alerts, and API’s CONSTANTLY during peak hours.
On a very busy day, last Friday, 6/1/12 (the day all the monthly reports for May are generated), this is what our 2 busiest databases looked like:
Note that the top of this graph is only 50% of total utilization, and we have a sustained, average peak utilization of just 15% of capacity. Click the graph for a VERY large version.
This is our second busiest database, with a little more detail on what the various “hills” in utilization mean (batch processing, monthly reports, etc.) — again, the top of the graph is only 50% of utilization:
We are barely scratching the surface of our database’s capacity. The same is the case for all our other systems which comprise the GPS Insight product.
(By the way, green means disk utilization, and red means memory utilization)
So what does this mean to our customers?
I just took a quick look today at just how many, and just how fast the typical query comes back during peak hours (10:30 PST is a peak time for us across the US):
We have systems in place to capture literally all of the nearly 100,000 queries per minute which hit these two busiest databases — here I simply choose the last 5 minutes to see everything which ran and run reports/statistics on it:
The data is loaded in just a few seconds:
And among the various reports I can run, we see what is called “service level” — this is how quickly we typically return the nearly 100,000 database queries (SQL) per minute which we receive throughout our system:
For these two database servers, we processed over 465,000 commands in 5 minutes, with an average response time of roughly 3.3 milliseconds (ms = thousandth of a second).
99.8% were taken care of in less than 200 ms.
That is a lot of processing for our customers’ benefit, and very fast, as well.
We keep our systems as responsive as possible in order to ensure to you that your vehicles’ locations, your API calls, and your reports and alerts will always be very fast and always available to you. Bear in mind each database server has at least one “twin” which is always in sync, should it ever experience a failure.
Also, please realize that we are processing 1550 operations per SECOND on these 2 servers at a 3 ms. average response time. That’s FAST.
Hopefully this helps you to understand what we do behind the scenes to manage your data and make it very fast for you to retrieve to help run your business.
Thanks,
Rob.
























