A10 Networks Thunder ADC Reviews

We use it to load-balance the website.

It's seamless.

The solution has made things easier in terms of operations efficiency It's one less thing to worry about. It just sits and it runs. 

We don't use many of the features. We're just using the basic ADC features. We're not really using anything particularly extensive on them. They sit and work most of the time. 

The ease of use is very good. It's very robust. It just sits and works. We forget that it's there a lot of the time.

The user interface is what people complain about most of the time, particularly if they don't use it very often. Then they complain that it's a bit clunky. It works from an ADC point of view, but the interface is a bit clunky.

We've been using it for about eight years.

It's rock solid. It just sits and works. That's the way you want it.

It's definitely scalable. We've not had any problems. I'm looking at the CPU graphs and it's not bothered at the moment.

The support is very good. They're very responsive if we ever have any problems. That was what drew us to them in the first place.

Prior to this one, we had another Thunder ADC. Before that, we didn't really have a requirement for an ADC. This was the first one we ever had. We've did have some free, software-based ones in the past. But when it became a bigger requirement, we ended up with ADC.

The initial setup was straightforward because we had an A10 engineer who came onsite and helped set it up for us. It wasn't plug-and-play. We did have to have some engagement. The deployment took a couple of days. We have continued adding more and more services onto it.

We put it in primarily for Exchange, to do some load-balancing at the time. These days, if we put it in, we'd have a lot more change-control to go through, but back in those days we just put it in, set it up, and away it went.

There aren't any licensing costs associated with it. It's just an appliance and you get all the licenses with it.

We looked at Kemp and there was another one, but it was about ten years ago. We went with A10 because of price and support.

Do research. I'd probably look at virtual appliances if I was going forward. One thing we could do with is a proper Dev and Test environment, which we don't have. I would have had some virtual appliances for Dev and Test. We did talk about that, but we haven't gotten around to doing it.

There are about ten of us who use it from a management point of view. But all the staff and students benefit from it.

It requires two or three people for maintenance.

We don't have any specific plans to increase usage. A lot of things might be going to the cloud, so there might be less use going forward.

I would give it an eight out of ten because it sits and works, it's robust. But the interface could do with a bit of work.

My primary use case is to use it as a software load balancer. 

Because of the industry that we operate in, we cannot use cloud. Therefore, we use it on-premise. We have 32 medium boxes, and that's what runs production.

A10 explained why the latency dropped significantly on a site that we have.

It gives me information on load balancing and offloading, which provides me information on performance.

I use SSL, TSL, and additional offloading. With additional offloading, this is where I can put my certificate on A10, as servers don't have the capability.

It has allowed us to smooth out our traffic.

The ease of use could have been created better. Some of the UI features are very primitive. Sometimes, wrong entries will go in and stay. I gave this feedback to the team who sold it to me. 

Also, some things like the traffic flow management take a lot of time to learn to use. While we have mastered using the feature, it doesn't tell us where we are going wrong or if something is breaking.

It will start failing if you have a containerized environment, which is why we have to start moving away from using the A10. If they could fix this, that would be good. 

It would be also be nice if someone could walk me through the solution’s Secure Service Mesh to optimize traffic within Kubernetes and containers, since I am not able to use A10 for rebalancing right now.

While I do use the security features on the load balancing, I find them to be primitive.

I have been using it for two and a half years.

I can't complain about its stability.

My engineering managers, with help of the DevOps team, manage the deployment. They have become self-sufficient and, with zero impact, they patch the servers in broad daylight, meaning mid-week.

I haven't needed to vertically scale beyond this box. It does scale horizontally.

The journey started with four servers and two websites and now I run close to 32 servers and service 36 applications, consisting of Web, application, and microservices. CPU utilization has moved marginally, by about three percent.

Troubleshooting always results in us contacting the tech support team and have a solutions expert, who is part of the sales team, send us helpful information. The solution expert is phenomenal, which is not the same experience that I have with the technical support.

I haven't been impressed with the technical support, though. When we need help, there is less support staff than we would like in the time that we need them. E.g., it may take an hour and a half to get assistance during an emergency. 

I would like them to provide learning tips and a community forum. where users can share ideas. They need more detailed support articles on the A10 website.

Before A10, I was using Network Load Balancing from Windows, which came pre-bundled and was primitive. I wanted to move to a professional version where I would have more control. I evaluated a couple of products and A10 seemed to be the clear winner. I had very simple parameters: the latency of response from our servers. The benchmark was NLB. When we compared the numbers with A10 and, when configured properly, it reduced the latency by roughly 15 percent.

There is absolutely no comparison between A10 and Windows load balancing.

The initial setup wasn't complex. A10 helped us with the setup. 

It took seven days to go into production, which was due to number of sites that we were running.

Implementation strategy:

1. It should have the ability, as a load balancer, to manage circuits.
2. Not everyone should have access to it.

I created a replica of the production box. Then, I set up the service, which I put on the load balancer. For 15 minutes, I moved the traffic from my NLB through A10 and monitored the performance. Then, I moved it back and it gave me the confidence that it could run safely. Then, I did the same for all 36 sites that I run. Afterwards, I moved it over to my team.

A10's solution expert helped us configure the box.

A third-party called Value Point helped us with the deployment. I was not impressed with their skills. 

For the deployment, there were two or three people from my team, two or three people coming from Value Point, and two from A10.

It's a phenomenal investment that the company has made.  

It improves our performance and management saving us money. 

There were budgetary constraints that keep us from investing in the single pane of glass traffic management feature. We saw a demo of this feature about a year to a year and a half ago.

We have not looked into any other options.

Go for it. 

It's always better to go with a nimble, growing company. Partner your growth with their journey. It's always beneficial for a stable setup. 

The primary use case is server load balancing.

We use the on-premise deployment model.

It has provided improvement in regards to flexibility of our applications between data centers.

It does improve operational efficiency, but it's not overwhelming.

The Global Server Load Balancing (GSLB) is simple to use.

It's not easy to use (mediocre at best). It's difficult for administrators to maneuver through the web user interface. It should be more intuitive through the web GUI. If you're a command line person, you can get around. However, the web UI is very difficult, and for our operational folks, they need the web UI.

The solution’s security features are minimal. I'm not impressed by the DDoS solution.

Traffic flow issues are very difficult, as there's no means for us to analyze the traffic coming in or out of the appliance without technical support. 

They need to improve in-depth diagnostics. I don't know how to do a tcpdump on the appliance. We need to do packet captures on the appliance to analyze what's going through it. Information is not as easy to attain as it is with other vendors. Better diagnostic tools would probably help.

I have been using the solution for three years.

Stability has been fine. It's been stable.

We upgraded our solution last year.

For deployment and maintenance, we only need a couple of administrators (less than 10 people).

The scalability is appropriate.

It's moving along based on the business growth, so there are no plans to increase usage.

The technical support is not great, but they are not poor either. They're fair. We don't open a support case that often, but when we do, it's not immediate as far as their response. When they do respond, they come up with a collaboration to troubleshoot or find an answer. So, it's fair at best.

The solution's support for our on-premise applications do their job. The basic functionalities for on-prem services are okay. They're not anything great. They do what they're supposed to do.

We previously had F5 and switched because of costs. 

We are able to do the same things (as the previous solution), but it is cheaper when we have to renew.

The initial setup was straightforward. The deployment took three to four months.

We were pretty familiar with the product, so we did the deployment ourselves.

It has broken even on ROI. We haven't lost any, which is good, but we haven't gained anything.

It is $7000 per unit for the support annually.

It does do the job, if your environment is simplistic enough. The product is fair for its market.

We test and used the single pane of glass traffic management, but we don't use that now. We went away from that. Now, we administer our devices individually.

The load balancing, proxies, and internal proxies are valuable features. 

• Improvement of the grid look
• Intuitive UI
• Syntax 
• Configuring

It is very stable. 

The scalability is really good. You can easily unify many devices at once. ADCs are simple to configure and each device is powerful. We have two engineers managing and monitoring multiple systems and it works well for us. 

For the past two and a half years, we have not had a need to open a tech support ticket. It is really stable. In the past, our experience with tech support was that they were extremely helpful. 

The setup depends on certain situations. In certain scenarios, it may be more complex than others. For example, while the initial configuration may be easy, the environment itself may be complex and that may limit the ease of deployment. It is easy for those who understand their environment.

The pricing and licensing are really easy because we bought the hardware and it has all of the license availability. The licensing is embedded in the hardware. We do not have to import any other license. The pricing is a third of the price of the F5 competition. 

We looked at A10 Networks Thunder ADC vs F5.

Carrier-grade NAT is the purpose of using A10 Networks Thunder ADC at Turk Telekom Mobile.

It solved the CGNAT performance issues and, within two years, it had no problems and no outages.

The Deterministic CGNAT feature.

I would like to see the scalability numbers for the virtualized version which are comparable to those I noted in my answer on "Scalability."

One to three years.

Every six months we made active-passive changes and upgraded to the latest version of the firmware. But we didn’t have stabiliy issues.

It scaled well for our numbers, up to 3 million subscribers for our most crowded region.

Technical support was mostly provided by another company, but A10 also responded our questions on time and accurately.

We used F5 load-balancers for the same purpose before, but those devices weren’t meant to be built for CGNAT. F5 devices built for CGNAT were also functioning well according to PoC tests.

It was easy. The CLI of the devices are universal and we had to make the solution as simple as possible in order to get scalability.

F5.

I would advise doing comprehensive PoC tests for your specific purpose, for all products.

One of the main advantages of A10’s line of ADCs is that they offer an all-inclusive licensing scheme for their products. This scheme enables enterprises to save significantly on costs, and makes managing licenses easier. A10’s ADCs is a solid, high-performance platform. It supports advanced traffic management and scripting features and broad set of security features using a third-party web application firewall. Lastly, A10’s ADCs have IPv6 support which gives it an edge over other offerings in the market should IPv6 functionality be required.

Though A10 has a solid offering, they have an on-going legal battle with Brocade. Depending on how this plays out eventually, this is a risk that must be evaluated carefully. Support can also be a problem if you are in the United States, as its main contact centers are in Asia. Application integration with third-party development tools is also a problem, as there are limited integration options available. Finally, A10’s products has limited web acceleration support for mobile users. This can be a problem if the application is intended for this purpose, but custom scripting is available to help alleviate this issue.

Overall, A10’s ADCs are a good product with the price many companies can afford and have features that even some high-end ADCs. I recommend it for those companies looking for advanced features offered by the product yet wants to keep costs down. However, the future might not be bright for this company, and I highly suggest that existing and potential users follow the progress of their legal battle with Brocade. Long term support might suffer if they lose their case, which a cause of concern for both existing and future customers.