Why Performance Vision ?
AUTOonline was looking for an alternative to our traditional network troubleshooting tools like Wireshark which were not providing great results in terms of Mean Time to Resolution (MTTR).
We needed to implement some form of continuous monitoring to have a history of our traffic and performance and to be able to conduct ex-post analysis. We also required a visibility that would go beyond the network layer and specifically provide insights on the content of the application transactions.
We previously worked with Horus-Net, a German company specializing in network troubleshooting, and noticed that they used Performance Vision to support their investigations, so we decided to evaluate it.
The results evaluation went beyond our initial expectations.
How do you use it?
AUTOonline is a brokerage company which facilitates the sale of damaged cars between insurance companies and recycling companies. It provides a complete software-as-a-service platform to share information between buyers and sellers, mostly in the form of web services and web applications. This platform is hosted in 2 physical data centers.
Performance Vision’s primary use cases are:
- Proactively checking the status of our services
- Enabling the handling of end-user complaints.
Proactive performance monitoring of our web services and applications
Performance Vision provides us with continuous statistics on the performance of our web services, This way, our team is permanently aware of who is using them, in what volumes, and whether the response times are normal or not.
In case of complaint from a customer, we can track back in time what happened, if there was an actual fault or degradation or a successful connection, and what was the cause of the problem.
For the web services, we capture the traffic between our load balancers and the web servers. PV continuously analyzes that traffic and gives us a permanent view of all the requests made. That way, we can view:
- All the HTTP pages consulted on our websites with their response times and eventual errors
- All the requests made to our web services with their response times, eventual errors and the details of the SOAP requests and responses.
In case of complaint or problem, we can share that information with our application team. We work in a DevOps mode; therefore, we deliver new features continuously. The visibility that PV provides helps us to resolve problems every week and has drastically accelerated our ticket-handling process.
Ad hoc performance troubleshooting
For specific needs, we extend the scope of the traffic analyzed by PV to other services like databases and file servers. When needed, we perform analysis campaigns on our database flows to determine whether the database service provided by MSSQL and Oracle DBs is slower than expected, if there are abnormal query volumes or specific transactions with slow response or errors.
While performing diagnostics, we also leveraged PV’s ability to monitor name resolution performances. In one specifically complex case, we would have suspected all the layers of our infrastructure, except the DNS architecture. With PV, we discovered our DNS response times had boomed from 20ms to 500ms.
We tracked back the source of the problem to a change made when AUTOonline joined the Solera Group:
- There were over 20 domain suffixes and the AUTOonline was among the last ones.
- The Web Proxy Auto-Discovery was generating a huge volume of DNS requests (due to all actions made by VPN users generating a DNS query).
The conjunction of these two events drastically slowed down the name resolution. It would have been extremely complex to diagnose this using traditional network tracing tools. Thanks to PV, we fixed this in a matter of hours.
Managing changes and migrations
In 2015, when we had to prepare our data center migration from Germany to the group’s datacenter in Switzerland, we identified potential risks of outages if some communications required for our systems to perform properly would not be enabled once migrated.
We had to compensate for a lack of documentation on old internal applications.
PV helped track all the communications from these systems and reverse engineer the applications so that we can set the firewall and infrastructure to let the necessary communications pass through.
Results and ROI?
PV has created huge benefits for our team and has been one of the success factors which made managing a critical and complex IT infrastructure with limited resources possible. It helped shorten MTTR and facilitated the collaboration between our Application and Infrastructure teams in a DevOps environment.
How do you plan on using it in the future?
We have other significant migrations planned in the coming months. We count on PV to make these changes smooth and without any impact on Audatex AUTOonline’s business.
SHARE THIS CASE STUDY
Getting started with a 15 day free trial today!
No credit card required.