A10 Thunder Hybrid Virtual Appliances [EOL] Reviews

A lot of what we use it for is application load balancing for certain applications. For example, with all the software development tool sets we use to load balance into the lash and tool sets that they're doing the software builds on. Often, they are also being used for VIPs for A10. It's on the A10, but it's also for active directory load balancing. We have some VIPs defined for active directory in theaters. Then that load balances to real VIPs behind it. There is also all our public facing stuff which also has A10 load balancing services provisioned on it, as well. We have public web pages that are being load balanced a lot of times onto the A10 onto the real servers and we also have logging load balancing going on, where just logged data is being sent to the VIP and the VIP load balances to multiple real servers behind it.

We have 15 A10s discovered in our environment right now, and some of them are spread all over the world. The management system is seen as AX 5630 in machine type.

The hardware versions are also seen as AX 5630. And then some of the ones down in Dallas at our DR sites are still the XA 5630s, but they're the smaller box.

In terms of most valuable features, I really wouldn't be able to give you one because it's a package deal.

On the A10 side, it was already here to begin with, so we had a shortened learning curve to stay with it. When Keysight split from Agilent, they already had A10s evaluated and pulled into the environment. At that time they had a lot of virtual context established for different business units. Once we moved into our refresh, we actually flattened that out some and did not do that as much because, we're mostly not putting in different application groups as owners into certain parts of the box. It's not something that we moved forward with. I'm pretty certain that's a popular feature. If the products are being pushed into a true service provider space where they're provisioning VIPs for multiple customers, I think that would be where each customer gets administrative access to their own VIP. I'm pretty certain AWS probably use the heck out of that feature, too. It was spinning stuff up.

In terms of what could be improved, when you look at the visibility of what is going on in the history, you have to go back through the logs. But these boxes are so consecutively busy all the time and the level of logging is so robust, sometimes it's hard to find things directly in the box and you have to go back to the log server to go find out what a session was.
Sometimes we have our AD team having problems and they're looking for a specific real host, a real client, that it's coming from, and they don't see it because we're doing a network address translation from the load balancer to the real server and all they see is the load balancer. In order to correlate that back to a real client, we need to go back to the log servers, then go and find it.

The reporting piece of that hasn't been 100% easy to do because of its volume and that it gets really large.

I have been using the virtual version of A10 Thunder Hybrid Virtual Appliances for about three years. On the hardware versions, we've had A10s in our network all the way back to 2014.

The A10s just got refreshed last year. We're using the VH Thunder for the Virtual Appliances. I think the one gig versions are the ones we are running with.

Stability has been strong. We have not had any issues with stability. 

There have been many times when we're doing packet captures, trying to find things, and trying to find out where the problem is, and if there was a tool out there that could easily see where the application performance problem is, that would be way more interesting in some aspects. All the data is there, but to sift through the volume of data and get to that level of detail of what the problem really is, is sometimes a challenge. More than once they say, "Well, it's network." So you dig and dig and you find out, that it's not really the network. It seems like eight out of 10 problems fall into that category. It is not easy to figure out at times and to find out what and where the problem is.

So it could provide a more accurate analysis of the performance problems.

It has scalability provided you get it sized right. I haven't seen any real performance issues with it. Anytime you're dealing with applications, you're putting another box in the path that adds complexity to that path. So, it would be nice to have better details on where the problems really broke when you're troubleshooting application performance.

When you look at the tech support of A10, they are probably comparable to any of these vendors' tech support nowadays.

It was pretty straightforward to set up.

We had looked at Kemp, too. We don't always just stay on one technology and a lot of times we look at things just to see where the market is going and things like that.

We see if we need more capacity or if there is a better service offering on the other side.

A10 Thunder Hybrid Virtual Appliances is a very workable box. I don't think any of these load balancers are overly discoverable in network management boxes. From what I've seen, all of them are always lacking on integration with MIPs.

On a scale of one to ten, there are no 10s. I would probably give a seven or an eight. When you look at A10 in general, they're known for being higher performance with less costs, and they're not going to be as security focused as an F5. Also, their flat pricing structure is good. You don't have to go back and spend money on every license when you enable new features. That's kind of a known thing with the F5, and it was a very big downer when you listened to that spiel. On the other hand, for people who are really into security, the F5 is better than the A10s. If you just want load balancers to function and work as infrastructure, then they function and work quite well.