Let us introduce ourselves. We founded Solecular with the vision of vastly improving costs and use of resources in IT, by creating an AI solution called Maxwell.
Just about any IT infrastructure, ranging from one rack to one million servers, is underused, meaning that even at peak usage, servers sit idle. For good reason, but at a steep cost. That’s where we plan to help.
With today’s tools, the most cost-effective approach is to overprovision, setting up far more overall capacity than can ever be used. This is true for containers or VMs, systems, racks and whole data centers. The result, however, is staggering. Steven J. Vaughan-Nichols of ZDNet states that up to 85 percent of the world's computing capacity is sitting unused in data centers. Scott Walker of Ericsson says that CPU utilization rates for non-virtualized servers, is about 6 to 15 percent, and virtualized servers can increase that average to around 30 percent.
Our discussions with early customers and our measurements in pilots confirm the same statistics. The image I’ve included at the top of this post is a screenshot of our utilization forecasts of a system in one of our pilots, which Maxwell derived based on our measurements over time.
We're building a number of tools to help you “scale in”, or grow without new costs.
Here's an example. If 80% of resources remain unused, even accounting for peak usage periods, theoretically, you could multiply your capacity by five without spending another dime. To be extremely conservative, we are confident that you could double your capacity with zero risk.
We know that uptime, performance, and flexibility take priority over costs.
- Top priority will always be to accommodate peak usage requirements.
- IT teams must make room for new workloads, either planned or emergency.
- When looking at efficiency, one must account for the cost of labor involved with improvements.
And yet there's still an opportunity.
At Solecular, the three priorities are always top of mind. We deliver a number of ways to satisfy all three and yet save significant amounts of money and waste by improving utilization.
Upcoming posts will talk about how recent advancements in AI and machine learning make this possible. We will cover how Solecular will begin by focussing on workloads, then systems, and then edge computing. We will look at the return on investment that comes with utilization improvements. And we’ll have some fun, introducing you to our team.
Thank you for visiting.