If one peer fails and corrupts the disk, will the other peer take over?

Yes! HAAst would start without any problem.

A lot of people don’t realize that all other PBX HA solutions in the marketplace are based on DRBD, an open source product which creates a mirrored logical disk between peers. As a result, any corruption to files on one system are immediately mirrored to the other system! This is why critical call centers like 911/PSAP’s, hospitals, etc. will not allow DRBD based PBX solutions – each peer must be fully independent of the other.

HAAst is the only product which does not mirror the hard disk. HAAst synchronizes individual files/directories/databases that you specify, and, it only synchronizes if the peer is healthy! So a failing peer will never corrupt files on the other peer.

You’ll find that many configuration generators now offer an ‘HA’ module, and they are all based on DRBD. Some organizations like Elastix are very honest about this – and show you how to install the open source product for free. Some organizations are less forthcoming, and actually charge you $3000 for a free open source product (DRBD).

Vendors of other PBX HA solutions are not telling you the whole picture. Many HAAst clients start with HA modules built into configuration generators, and switch to HAAst after their ‘cluster’ couldn’t withstand any failure other than the most simplistic scenario of unplugging one box.

There are lots of other differences too. DRBD solutions don’t allow for configuration differences between peers. HAAst allows peers to have different trunks, users, dialplans, etc. (yet still share a common base that is synchronized), all things you can’t do if you simply mirror a disk.

Telium is a big supporter of free and open source software, and DRBD is a wonderful product. But adding DRBD to a PBX and calling it ‘HA’ is naive; just like adding RAID to a PBX and declaring it ‘HA’ is naive.