I like to work on old motorcycles (touring bikes from Japan). And I typically have a few projects going on at the same time. A lot of projects means a lot of parts and a lot of different tools. They are everywhere in the garage. A bit of a mess, really
Bike projects always seem to take longer than I think they should. A few hours turn into a few days, then weeks or longer. Why is that? It’s because often I can’t find the part, tool or diagram I need to finish the project. Yes, the shop is messy but it is the number of projects and parts that makes it slow going.
What does working on old bikes have to do with faster business analytics? Simple. I know I my shop will always be busy, but how much more could I accomplish if I could find what I need faster? What if instead of taking me half an hour, it took just a few minutes?
It’s like looking for essential, actionable nuggets in heaps of business data. You’re pretty sure you know they are in there, somewhere. You just need to find them faster so you can understand them better and make more informed decisions.
Our new Micron 5210 SSD can help you query your business data faster (ten times faster!) and find those nuggets to make more informed decisions sooner.
First Things First: Speed Matters
Since datasheets may not tell performance stories at a platform level, we measured storage throughput for a (relatively) speedy HDD configuration (10K RPM HDDs) and a much speedier Micron 5210 SSD configuration.
The difference? The Micron 5210 SSD platform showed 8.6x higher storage throughput.
That’s a pretty good start to understanding the differences we expected to find in query processing (next section).
These results aren’t simple, single drive tests – this data is for the complete, multi-drive configuration. Same host system, same CPUs, same DRAM.
The only difference was the storage.
Finding Nuggets Faster: Data Analytics Processing
While the throughput looked great, we wanted double-click data: what would that 8.6x throughput difference mean in terms of query processing? Turns out, quite a bit (spoiler alert: more than 8.6x!). Here’s why this metrics is so imporant: for analytics systems, the most important metric is the time it takes for the system to complete a known stream of queries (think of it as stream run time). When stream run time is lower, the queries process faster.
This is what enables faster, more complete answers: the abilty to query faster so you can query interatively (run additional queries in the same timeframe). And the Micron 5210 SSD is the right tool for this job. Take a look at single stream, relative completion time below. The Micron 5210 SSD is on the left, the legacy HDD on the right (remember, these are 10K hybrid HDDs). Naturally, lower elapsed time is better:
Stream Run Time (lower is better)
Is Query Speed All There Is? What Else Is Important?
Our quest to find and use locked up data value, the need for fast analytics and enhanced in-place data analytics drives new ways of thinking about storage. But we know that we must balance this demand with responsibility – we need to be aware of and, more importantly, efficient in our use of, resources.
Energy-efficiency: These are precious resources so we should be stingy. The less of them we use the better. The Micron 5210 SSD proved energy-stingy in our testing (expressed as KWh to complete a standard query set).
Approachable Price Points: Financial resources are precious too, and with four bits per NAND cell, more applications and workloads can move to SSDs more affordably with the Micron 5210 SSD. There’s more data, configuration and test details in this Technical Brief: Three Paths to Better Business Decisions.
You can also check out Micron.com/5210.
What’s Next for Your Analytics?
The way we approach analytics changes as we understand how we can better choose and tune our technology. Sometimes these changes are evolutionary, sometimes they are revolutionary. For years we’ve used hard disk drives (HDDs) for our business intelligence (analytics) systems. When we wanted faster performance1 we used faster RPM HDDs (an evolutionary approach).
We found that revolutionary change often brings amazing results. Our Micron 5210 SSD combined with business analytics is one example of this: faster throughput means faster analytics that consume less power.
I think there’s a lesson in here for my next bike project – greater efficiency and greater speed go hand in hand. Sounds like I need to rethink my own workflow in the shop, doesn’t it?
How do you manage business intelligence today? What do your platforms look like, and what would you like them to look like in the future? What benefits could you see from faster business queries with higher energy efficiency? I’d like to know. Reach out via Twitter @GreyHairStorage.