Truth, lies and reverse incremental backups.
Updated November 11, 2015
Truth, lies and reverse incremental backups.
Score 1 out of 10
Overall Satisfaction with Axcient
As a managed services provider, we've deployed Axcient units to dozens of our customers environments to handle backing up their entire server infrastructure.
- File-level restore speed. Because of the way backup data is stored (most recent backup is stored as a full image, with incrementals available for going back in time), restoring recently lost data is reasonably fast. This is, most often, what we find ourselves needing to do so it's a nice attribute. It's also the inverse of how several competing products store their data.
- Easy to deploy. Getting the units on-site and on-line is plenty easy enough that it can be delegated to the most Jr. Level field tech and they'd be hard pressed to do it wrong. Once the unit has an internet connection, all provisioning can be done remotely - and even then, there isn't much to it.
- We have a 100% failure rate for full-system restores. After our 3rd complete failure, Axcient flew an engineer to our location, they brought along their own BDR with a copy of our off-site data and spent two full days attempting to restore both from the BDR they brought and the BDR in the customer environment - they went home a complete failure.
- Axcient support is singularly focused on supporting their appliance, regardless of consequences to the rest of your environment. They have no problem recommending such things as deleting all shadow copies on a protected system when their software fails to clean up the one that it has created. When investigating issues I have reported to them they have repeatedly kicked off full backups of systems in the middle of the day and let them run until users complain about system performance. They continue to do this despite my increasingly bitter complaints.
- Linux "support" is a significant overstatement. Linux systems can only be backed up through Samba. You cannot do a "full system restore" of a Linux machine, nor can you run it as a virtual machine on the appliance. Being backed up through Samba means file attributes (ex: immutable, append-only, etc.) are lost and only "regular" files can be backed up (forget about device files, fifos, sockets,etc - all of those would need to be recreated manually in a disaster recovery situation). Axcient has no capacity to run pre/post backup commands on the target system, so you cannot directly trigger `flush` or mysqldump or anything else useful in making sure you're getting anything other than a crash-consistent backup of some selected files. Don't fail to realize that this backup method means you'll have to log into your Linux machine through Samba as a user with read permission on all files you intend to back up; Axcient recommends you use "root" (a cardinal sin).
- Backup times are mysterious and impossible to predict. Examples from a single customer environment: A server with 620GB of data (and a high change rate) takes ~6 hours to back up and a server with roughly 90GB of data (and a particularly LOW change rate) takes 9.5 hours. The times for these machines are consistent. These are both virtual machines running on top of ESXi 5.1 with with more than enough RAM,CPU and IOPS dedicated to them. Axcient support simply states that this is normal behavior and we are not to be concerned.
- Everything is slow. 9.5 hours to back up 90GB of data. Offsite transfers consume an inordinate amount of bandwidth compared to competing products we have deployed. My suspicion is that this is at least partially because entire files are being copied instead of changed blocks.
- "Image level" backups aren't actually image level at all. All backups are done at a file level, the "image" backup job type just means it will automatically select all volumes on the target system and get a system state backup as well.
- "Agentless" backup system... has an agent. Scripts and a delightful executable by the name of "axexesvc.exe"are pushed to and executed on each protected windows system for every backup. This "agentless agent" can have all of the same problems that any other agent can have (such as hanging, running forever, consuming all available system resources, etc.), but also brings with it a fairly unique added gaping security flaw. axexesvc.exe is effectively a trojan horse from Axcient, allowing Axcient's employees to remotely execute any arbitrary command or code on any system in your network using the privileges of whatever account you run your backups as (commonly and per Axcient recommendations: a domain admin level user). There are no restrictions on what they can do or when they can do it and there is no logging of their activity. We have had multiple occasions where Axcient's support staff have abused this "back door" to the detriment of our customers production servers.
- The RMC provided by Axcient has made for seemingly meaningful (though admittedly not directly measured) efficiency gains in monitoring backups for a couple dozen different customer environments, but the time spent running down failures and endless conference calls with Axcient executives more than eat up this perceived efficiency gain.
Axcient has two things over on Unitrends. An MSP friendly console for managing several devices (though I'm told Unitrends will have this very soon), and cost. Unitrends does proper block level backups of a huge variety of operating systems (that it can actually restore). Can run as a virtual appliance (which Axcient will have soon) or a physical appliance. Is incredibly fast compared to Axcient and vastly more flexible.
- Backup Exec
In terms of a platform for backing up and restoring data, Backup Exec is tried and true and incredibly reliable. In terms of complexity of management, it's somewhat like a space shuttle (Backup exec) vs. an easy bake oven (Axcient).
From a purely technical perspective, I'd take Axcient any day over the non stop problems we had with Zenith. At least Axcient actually has data on it when it says a job has completed. Virtualization component in Axcient also actually works which is a plus.
- Barracuda Cloud backup
On par with Zenith
If you have or will have any Linux systems in your target environment, Axcient is not a proper fit. If you have backup window or bandwidth constraints, Axcient is not a good fit. If you have more than about six servers or a total of 1TB worth of data, Axcient is probably not the best fit. If you are subject to any kind of regulation (ex: HIPPA, SOX, CJIS) and don't enjoy the thought of Axcient support exploiting domain admin privileges on your servers to execute arbitrary changes, run screaming in any direction other than Axcient. If you have a small office of around 3 Windows servers with a relatively low rate of change and a modest recovery time objective, with a decent upstream connection that it can use unencumbered through the night and zero interest in the integrity or privacy of your data... Axcient is probably a pretty great fit for its ease of deployment and ease of management. In the actual words of one Axcient VP "It probably makes sense for you to run Unitrends while we are validating things"
Issues with failing backups drag out for months at a time. Axcient hardware replacements are quick to be suggested by support but as the issues are most often with the software architecture this almost never helps. Support often uses their access to the appliance as a back door into customer environments without our consent and despite our repeated complaints, accessing our customers servers and executing disruptive diagnostics like chkdsk in the middle of production hours for that customer environment. More recently they've been blaming their frequent off-site transfer failures on our customer's firewalls (of all makes and models) and insisting that the only possible resolution is to increase the TCP session timeout dangerously high (multiple hours), putting every client environment at substantial disk of even accidental Denial of Service attacks. Talking to any two support individuals, even when escalated all the way to their development team, often yields entirely different and contradictory answers regardless of the problem.
Problems left unsolved
Not kept informed
Difficult to get immediate help
Need to explain problems multiple times
Support doesn't seem to care