The Myth of "Guaranteed" Resources
If you are running a high-traffic Magento store or a critical MySQL cluster on a standard budget VPS, you have likely experienced the "Phantom Lag." Your load averages are low, your memory has free space, yet your page load times just spiked to 3 seconds. Why? Because your neighbor on the physical node decided to compile a kernel or run a backup script.
In the current hosting market, OpenVZ and Virtuozzo dominate because they allow providers to oversell resources. They share the host kernel. If one container panics, the node wobbles. For a hobby blog, that's fine. For a business targeting the Norwegian market where uptime is currency, it is unacceptable. This is why at CoolVDS, we have standardized on KVM (Kernel-based Virtual Machine).
KVM vs. The Rest: True Hardware Virtualization
KVM isn't just a container; it's a hypervisor that turns the Linux kernel into a hypervisor itself. It was merged into the Linux kernel mainline back in 2.6.20, and with Red Hat Enterprise Linux 6 (RHEL 6) shifting focus to KVM, the writing is on the wall for older Xen implementations.
Unlike shared-kernel solutions, KVM allows you to run your own kernel. Need to enable specific modules for `iptables` or tune your TCP stack for low latency without begging support? You can.
Pro Tip: To verify your server supports hardware virtualization (required for KVM), check your CPU flags:
grep -E 'svm|vmx' /proc/cpuinfo
The I/O Bottleneck: VirtIO Drivers
The biggest criticism of full virtualization used to be I/O overhead. Emulating an IDE disk controller is slow. In 2011, we don't emulate; we paravirtualize.
By using VirtIO drivers, the guest OS knows it's virtualized and talks directly to the hypervisor. This reduces syscall overhead significantly. When configuring your KVM instance (or asking us to do it), ensuring the disk bus is set to `virtio` rather than `ide` can double your disk throughput.
Optimizing Disk Access for Database Loads
If you are running MySQL 5.1 or 5.5 on a virtualized EXT4 partition, the access time updates can kill your I/O performance. Here is a quick win for your `/etc/fstab`:
/dev/vda1 / ext4 defaults,noatime,barrier=0 1 1
Setting noatime stops the filesystem from writing to the disk every time a file is read. It’s a small change that saves thousands of writes per second on a busy server.
Storage: The Move to SSD
While 15k RPM SAS drives in RAID-10 are the industry standard for enterprise stability, the emergence of Solid State Drives (SSD) is changing the latency game. For database-heavy applications, random read/write speeds on SSDs are orders of magnitude faster than spinning rust.
At CoolVDS, we are beginning to roll out SSD-cached storage tiers. While expensive, the IOPS (Input/Output Operations Per Second) gain is undeniable for datasets that don't fit entirely in RAM.
Data Sovereignty in Norway
Hosting physically in Norway isn't just about ping times to Oslo (though dropping from 40ms to 4ms is nice). It's about the Personopplysningsloven (Personal Data Act). The Data Inspectorate (Datatilsynet) is increasingly vigilant about where sensitive data lives.
By keeping your KVM instances within Norwegian borders, you simplify compliance with the EU Data Protection Directive. You know exactly where your bits are physically located—not in a vague "cloud" floating somewhere over the Atlantic.
The CoolVDS Implementation
We don't believe in overselling. When you buy a 2GB RAM KVM slice from us, that RAM is reserved for you. It's not "burstable"—it is yours. We use KVM to enforce hard limits on CPU and RAM, ensuring that a noisy neighbor can never steal your clock cycles.
Combined with our carrier-neutral upstream connections ensuring low latency across Scandinavia, KVM provides the stability required for serious systems administration.
Ready to ditch the lag? Stop sharing your kernel. Deploy a true KVM instance on CoolVDS today and experience the stability of dedicated resources.