OnApp is a software suite for managing your own HA cloud infrastructure. It was originally created by AtechMedia, a UK-based software house based in Dorset and called Radar.
This was sold some years ago and whilst the core of OnApp has likely been completely re-written there are still some core principles at use within the software.
We were an early adopter of OnApp, on their $10 per core pricing. It was their base offering, at the time they didn’t have any additional ways to relieve you of your profit, with services like CDN or storage.
Coming from 3Tera AppLogic which was going through its own challenges (due to being acquired by CA), we shifted our data. As is common, this is really the only time OnApp went out of their way to help; migration of data.
Although OnApp wasn’t on v1 release (when we started using it), we’d not be considered early adopters but based on how unreliable and buggy the software could be, you’d be forgiven for thinking otherwise.
The problems over the years ranged from:
- Failing backups, resulting in data loss
- Dual running virtual machines on different HVs, resulting in data loss
- Support deleting virtual machine data “accidentally”
- Upgrades leaving the cloud in a worse state than before the upgrade
- Fixes for newly introduced bugs would take months to fix
Although not an exhaustive list, two main and consistent problems remained until the point we moved away; poor support and lack of taking responsibility.
The support is mainly in the US and Ukraine. For first line issues, most is dealt with in the US. For more technical issues, it goes to Ukraine. The response time from being passed from US to Ukraine could sometimes be in days, often needing t be chased by phone, or even a new ticket. The Ukraine team had a habit of saying “this has been fixed, please check” without giving any debug on what the issue was, or how we could fix it.
Often they would just go ahead and make changes resulting in downtime for virtual servers because the controller would then reboot a HV as it believed it was down, or in the worse case for us, completely deleting a virtual machines LVM storage by accident.
We raised our concerns multiple times and largely nothing happened and a quick search would attest that we were not alone.
Towards the end of us using OnApp, our environments were upgraded. We always had OnApp do this as it was included in our licence cost and it ensured they checked everything. However, that appears to not be the case. On our last upgrade, this broke several of our HVs causing them to kernel panic. Our hardware was blamed, the setup was blamed, our settings were blamed; I could go on. They failed to understand that our hardware was approved when it was deployed (and still was at that date); when they quoted some settings were wrong, they ignored the fact they had set the HVs up as part of their initial configuration and each time they quoted their terms and hid behind the usual “hardware is not our problem” response – without considering that their upgrade had caused this issue to appear on hardware that had been happy and in place for 2+ years prior.
For software that is designed and sold on the basis of HA, we had enough. On switching to the alternative offering, the same hardware worked flawlessly and the HVs have been rock solid since; funny considering the hardware was blamed and we’re still using the same virtualisation technology (KVM).
Simply, they’re not worth what they charged at the time and we’re aware they have increased their minimums. Whilst they tried to screw us on our licensing, they quickly backed down and blamed that on another “human error”.
There are other better paid-for and for-free options out there if you’ve got the skills to integrate and use more command line and API only tools.