High Availability for Asterisk

Compare HAast to alternatives

HAAST

»

COMPARISON

Overview
Benefits
Features
Technology
Testimonials
Screenshots
Videos
Editions
FAQ
Comparison
Download
Pricing / Buy
Forums
License

comparison

 
HAast
Product A
Product B
Product C
Product D

Peer Autonomy

No shared physical devices
No shared logical devices
Support unique peer configurations
Critical resource isolation

Synchronization

Config, data, and voicemail files
Asterisk® internal database
No sync if detect peer unhealthy
No sync if detect corrupt data
Intelligent merging of data after reconnection

Peer Health / Failure Detection

Asterisk® process awareness
Intelligent Asterisk® testing
Hardware health awareness
External environment awareness
User defined health scoring/rules

Geographic Separation

Capable of local (inter-data center) separation
Capable of wide (intercontinental) separation
Automatic WAN network latency compensation
Adjustable LAN/WAN network latency control

No Single Point Of Failure

No imposed shared utility/communications
No dependance on shared physical device
No dependance on shared logical device

Encryption

Encrypted control traffic
Encrypted data exchange traffic
Encrypted file storage network traffic
Yes/Excellent
Partial/Fair
No/Poor

Comparison Details

The following boxes provide further comparison details, and by clicking the down arrow at the bottom of each section you can see how HAAst performed.

Peer Autonomy

Peer autonomy describes the degree of independence and isolation of peers. Peers in a cluster should be fully autonomous, not sharing any logical or physical devices. If peers share any logical or physical devices then failure of the shared device immediately shuts down the entire cluster. Clusters must also allow for differences in peers (including hardware, OS, drivers, network, and Asterisk configuration) to ensure peers can operate in dissimilar data centers while still sharing configuration data.

HAast is the only high availability product which does not share any logical or physical devices between peers. All other products share USB-based trunk/channel banks (e.g. Xorcom TwinStar, Digium R-Series banks), and/or storage (e.g. NFS, SMB, DRBD). Failure of a shared USB based channel bank immediately causes the entire cluster to fail. Failure of shared storage, or corruption of shared storage, immediately corrupts / disables both peers.

HAast is also the only high availability product which allows for minor and major configuration differences between peers (eg: VoIP trunks vs T1 trunks), applying changes to common configurations following synchronization. This allows for cluster members to operate in different data centers, with different trunks, different routes, etc.

Syncrhonization

Synchronization describes how well peers keep critical files and databases in sync, while also avoiding synchronization of corrupt files caused by a failing peer. If peers share file storage (eg: NFS, SMB, iSCSI, DRBD) then a failing peer may corrupt files needed by the healthy peer, resulting in immediate cluster failure.

HAast is the only high availability product which does not use a shared storage device (e.g. NFS, SMB, DRBD or iSCSI). HAast is also the only product to synchronize files / databases only when both peers are confirmed healthy, and only from the active to the standby peer. All other products immediately replicate corrupt files in the event of either peer failure.

Alternative products use shared network storage which is not sustainable over a WAN due to network traffic, or worse, use DRBD to share a logical disk which can become broken (“split brain”) after peers are separated. These solutions risk corruption of data on peer reconnection, loss of voicemails/configuration data, etc. HAast on the other hand applies changes only in one direction, and resolves post-reconnection conflicts based on which peer was active prior to, and during disconnection. HAast is also the only product guaranteed to maintain referential integrity of databases since there is no risk of disconnection of shared storage halfway through an update (HAast performs database-level transactions, not block-level copying of database files).

Peer Health / Failure Detection

Peers are considered healthy when they are able to provide PBX functionality to users. Peer health must consider not only simple factors such as the Asterisk process being alive, but whether Asterisk is bridging calls, whether the hardware is functioning properly, whether the firewall is functional, whether network routes and ITSP services are accessible, etc.

One competing product depends completely on Asterisk to announce when it has failed. Another product looks only at the existance of the Asterisk processes to declare a peer healthy. The third, the open source alternative, allows users to write scripts to test what they can from the command line, but by default has simplistic Asterisk awareness, and no environmental awareness.

HAAst uses a variety of built-in sensors which monitor hardware, software, the Asterisk process, Asterisk operations, network routes, SIP connectivity, etc. HAAst supports both predefined sensors and user-defined sensors, combined with sophisticated rules which allow for assessing health using administrator-defined formulas.

Geographic Separation

Geographic separation describes how far apart peers can be placed and continue to operate as a single cluster. Peers must be able to operate across separate and disparate data centers (backup/primary), and ideally across wider distances (eg: states / provinces / continents) to address the impact of regional power / datacom / telecom outages.

HAast is the only product which can work across data centers, whether in the same city or on different continents. HAast even automatically compensates for variable network latency, avoiding false positive detections due to lengthy and unpredictable network routes.  HAast is designed from the ground up to operate over a wide area network, and is not dependent on any LAN/proximity-based protocols/devices. HAast sends only file / database differences between peers, and compresses (and encrypts) all data communications.

All other products either share a USB device which requires peers be placed within 10 feet / 2 meters of one another, and/or use shared storage (e.g.: NFS, SMB, DRBD) requiring peers reside on the same local LAN (or consume enormous bandwidth). None of the alternative products can handle reconciliation of data on cluster reconnection.

No Single Point of Failure

No single point of failure refers to whether or not a cluster is susceptible to, or introduces, a single point of failure which can shut down the entire cluster. The “Single Points of Failure” criteria considers the necessity of shared physical/logical devices, devices which must be placed in front of the cluster (to route calls), requirements for peers to be placed in the same data center, etc.

HAast is the only product without a single point of failure, and is the only product compliant with NG911/PSAP (Next Generation 911/Public Service Answer Point) standards for emergency services call centers (with no single point of failure).

All other products share logical / physical devices, and must be located sufficiently close to one another to share power, datacom, and/or telecom networks.

Encryption

Encryption refers to the protection of any and all data which moves through the cluster, as well as to/from any shared devices. This includes voicemail contents, call logs, caller ID information, configuration data, cluster control commands, etc. Legal requirements (e.g. HIPAA, PHIPA, NG911, Sarbanes Oxley if applicable)  or security requirements (e.g. military, government) may demand a particular level/type of encryption of all such network traffic. Healthcare environments may also deem this data to be ePHI in transit, requiring encryption.

HAast is the only product which offers encryption for all traffic between peers.  All competing products use completely unencrypted communications, including plain text protocols (e.g.: DRBD, NFS) for writing to shared storage (including saving of voicemails). Use of plain text communications by all other products prohibits their use in health care and NG911 controlled environments, and may violate Sarbanes Oxley regulations depending on the nature of the business.

HAast uses 56-bit DES encryption in Free and Commercial editions, with 256-bit AES encryption available in OEM editions (available only to military, emergency service, and security partners as permitted by export control regulations).

Cluster Failure Detection vs. Recoverability