Home › Forums › HAast (High Availability for Asterisk) › General › HAast alternatives
Tagged: comparison, haast, quality, works
-
AuthorPosts
-
I’ve spent 3 months evaluating alternatives for HA for my company, and I have to say it’s a sad world out there. I found a few products that claim to create HA/cluster for Asterisk, but they are basically junk; and these are PAID commercial products. They are little more than simple scripts (in one case PHP) that start Asterisk if the other server disappears. I tried a couple of FREE approaches (for example using the “HA in 5 minutes” video on youtube). Utter nonsense.
In summary:
- They only detect if the other sever is gone. No “sensors”, no checking if Asterisk is bridging calls, no checking if the Asterisk process is working (just alive), etc. They are STUPID. In my tests they would not fail over even if Asterisk had locked up because they just provide a simplistic check.
- They either blindly copy (“sync”) everything, or they depend on a configuration generators to export/import settings. In my tests when one node crashed and was corrupted, they would “export” a backup to the server, which would then “import” a corrupt backup. Bottom line, one node going down destroys the other.
- They don’t rollback any sync’d data in case the active server is failing, in fact a couple of times they left the standby server in a non-startable state because they copied over corrupt configuration when the primary was failing.
- They would cause unnecessary failovers if there was even a little network latency (false positive), or not failover when the other server was corrupt (false negative)
- They create multiple points of failure. Instead of improving availability, they create a complex collection of discrete servers to wrap around Asterisk, but if any one of THEM fail then the cluster dies. Ridiculous! Now you have 4 single points of failure, and each of those need redudancy.
HAast was the only product we tested that actually works! We ran one server in the AWS cloud (Ireland) and one local (in Virginia). Passed with flying colors! None of the above problem, and HAast successfully kept our call center up throughout the tests. I would be happy to share details or post our full test suite results if anyone wants.
I’m posting this 1. to say thank you for creating a real product, not trying to take my money for a piece of garbage, and 2. to advise others to test – don’t trust vendor claims. All the products worked great in the simple “unplug one server” test. But when it came down to real world failure scenarios only HAast actually worked. Seriously, I can’t believe what some companies charge for their “HA” when they offer little more than a 10 line PHP script.
Don’t trust online claims! Every vendor I spoke to said their product is the same as HAast. It’s sort of like bicycle vendors telling you their product is the same as a truck vendor’s product. Sure, they all have seats, lights, wheels, seating….but they are not the same. They are selling you a bike, telling you it’s a truck. If you are running a business with important phone system, test for yourself. Only HAast is actually what it claims to be.
Thank you! We have posted your message unedited for everyone to read.
We have been in business for 28 years, and have seen “HA” solutions come and go. There’s a reason that we are the solution trusted by 911 (PSAP) call centers, hospitals, fire departments, military, and more.
A significant part of our business is customers who initially went with cheap/DIY solutions. Those solutions too looked great in simplistic tests. But after a few real world HA failures (non-recovery of communication services) they switched to HAast.
-
AuthorPosts
- You must be logged in to reply to this topic.