What is BetterLinux? The Only Way to Get Both Fixed and Auto-scaled Resource Limits.
BetterLinux runs on popular Linux distributions to optimize shared servers for hosting companies (as well as optimize servers for data centers, SaaS, and cloud environments). It enables hosts to control the allocation of resources to individual users. It controls CPU, disk I/O, processes, memory, MySQL, and network bandwidth resources, all within a secured system.
Proper use of BetterLinux will vastly increase the number of customers you can fit on a shared server with no decrease in system performance or added costs. It stabilizes systems, reduces outages, runs smooth under stress and attack, and guards users from each other, greatly reducing support tickets and administrative labor.
Web hosts have access to only two resource-limiting methods: 1) Fixed or Static limits, the traditional approach, and 2) Dynamic or Auto-scaled limits, a newer and more advanced technology. Both methods are valid, and each has advantages and disadvantages. BetterLinux is the only product to offer both.
Fixed limits, what most are familiar with, make it easy for customers to do capacity planning for their websites. A website owner runs a benchmark with fixed limits, gets numbers, and can expect those numbers to remain fairly constant over time—no more, no less. The downside of fixed limits is that it is essential to avoid putting too many users on a shared server. If you do, you risk bad performance, or worse, taking away resources from critical system operations so that not even admins can save the server.
What many do not consider with fixed limits is how hard it is to choose the right limits to avoid over-commitment. Take I/O, for example. Many do not have the same disks in all their servers—some have fast disks, some slow, some attached to SAN, etc. These heterogenous devices all have unique capacity, requiring unique user limits. Even with homogenous devices, it’s hard to determine physical disk limits in terms of iops or MB/s. There are similar problems choosing CPU limits. As a result, fixed limits require you to guess each server’s capacity, how many users to put on a box, and what their limits should be. Finding the optimal balance requires tedious trial and error. Even if you get it right, it will be right only temporarily until usage patterns change, which they always do. What is optimal for a typical workload will obviously not be for an atypical one. Atypical loads will happen. Knowing this, your choice is either to under-populate the server in anticipation of atypical high loads (and lose profits by having fewer customers on your server), or cater to a typical load and risk unprepared the effects of atypical high loads (and cause periodic slowness or downtime for your customers). In sum, fixed limits force you to choose between two undesirable options: Fewer profits on the one hand, or periodic poor performance on the other.
But those are not your only choices. Dynamic or auto-scaled limits is a newer technology designed to eliminate a forced choice between profits and performance. It also eliminates the difficulty of choosing optimal limits. This is because auto-scaling automatically finds new optimal limits every second as a function of the ever-changing system workload. When you turn on auto-scaling, if your fixed limits are too low (ie., some users are maxing out their limits while other system resources sit idle), auto-scaling raises all limits proportionately to create full system utilization and increased system-wide productivity. If the fixed limits are too high (ie., users aren’t reigned in enough and the disk is getting congested), auto-scaling lowers all limits proportionately to allow maximum usage while preserving resources for critical operations and administrative control. Atypical high loads are never a problem. Even on the most overcrowded server during the highest atypical load, there is never downtime, the system is always usable, critical operations are always protected, and admins can always issue commands. Simply put, auto-scaling fixes every difficulty and inefficiency of fixed limits. Fixed limits can never match the performance level of an auto-scaled shared server.
With BetterLinux, you can use either fixed limits or auto-scaled limits. The choice is yours. In fact, you can switch between these methods at any time in a matter of seconds without worry, even on a production server.
Can any BetterLinux competitor come even close to the customer density of a shared BetterLinux server? No, and here’s why. Atypical high loads are inevitable, and fixed limits are totally unequipped to handle them. If performance consistency matters to you, even during atypical high workloads, BetterLinux will double the shared server density of our closest competitor. That’s twice the profits for you. And as for individual website owners, they’ll never get less than they bargained for, but they’ll often get a lot more. Welcome to BetterLinux.