Our Servers Honk -- Thanks AMD

Our servers really honk.  The site is pretty fast.  No, it's faster than that.  And we currently (2014) are doing it for $80/month in hosting charges.

How'd we do it?  Wth the latest in high-quality hardware and software.

In recent years, there has been a transition in hardware and software that greatly improves database performance.  It started in 2003 wth some changes AMD (see our Advance Micro Devices, Inc board here:  AMD) made in how their CPUs accessed memory.  Intel quickly folllowed suit. And software that could take advantage of these new CPU features was also fairly quick to follow -- depending on what operating system you used.

These new CPUs were able to address a bigger memory space than the CPUs that came before them.  A much much much (keep going here) bigger memory space.  And, one of the primary benefactors of the change was the lowly database.  Using the new CPU hardware and the new software designed for it, databases became sometimes 10-50 times faster on standard select statements. And then we further boost performance with caching technology avaiilable only on open source platforms and with database caching available in our app layer.  Our server is nearly sleeping even at peak load now.

And then, another revolution hit -- the dual core and multi-core CPU came along.  Soon we had 2 core CPUs, then 4 core, then 8 core.  And the top of the line stuff today is well above an 8 core CPU.  This also helps things like databases particularly well. 

And then yet another revolution hit.  This time it was virtualization which allows you to run many copies of an operating system on a single machine. This allowed the cost of hosting to come way down while allowing better utilization of hardware.

And then yet one more revolution hit -- SSD drives.  SSD stands for 'solid state disk' drives and they replace the old spinning disks with a bank of FLASH memory chips.  Made possible by Moore's law which says the number of transistors on a chip doubles every 18 months or so.  That has lead to FLASH chips that have so much storage space that a small bank of them can replace a hard drive.  And chips are faster than mechanical spinning platters so now you have blazing fast bulk storage that also uses a lot less power than spinning disks.

We use all this technology on the IDB site -- the latest CPUs, operating systems, database software and SSD drives -- and it makes for a site that's limited only by Internet network bandwidth, not sever hardware or server software.  We got a massive speed boost when we moved to SSD drives and the SSD technlogy is just now making it into enterprise-level products and tools meaning it has a lot or room left to run in terms of getting faster and bigger between 2014 and 2016 years.

As the IDB site grows, our hardware will scale with us and keep the site fast as it grows.

Need more database power? We have multiple ways to do that including adding db mirrors.  We can do all the main upgrades without taking the site down including adding DB mirrors, adding webheads, and tweaks to stuff that's beyond cool but also super geeky (I could go on here for a while).  

We can add and remove resources from each layer of our stack, as needed.  We're not limited to just jumping from one hardware implementation to the next bigger one. Rather, we have a lot of granular options to control how we move up the capacity scale.  All in the name of a blazing fast site that scales with the user base and database footprint.

The last aspect of speed is server location.  We selected New Jersey and Texas to locate production servers. We also have off-site backup infrastructure in Michigan.  This gives our production environment a dual-homed presence to assure uptime and speed to all markets. 

We know speed is important.  Not all sites are fast, many reasons why, but on the 'burns vs. blows' scale, we hope you find that IDB is a burner!!  Always.

Enjoy the site!

The IDB Team