Editor’s note: The retailer featured in this post has given us permission to tell their story to help their fellow IT Operations teams solve similar challenges. Their brand name has been redacted for competitive and privacy reasons.
The Retailer’s Business Challenge: Lack of visibility into the performance of their complex environment
A top 5 global retailer uses Citrix Platinum Edition SDX NetScaler Application Delivery Controllers to reverse proxy their main retail website. The problem, however, was that management faced a lack of visibility into the responsiveness of the site to information requests and transaction performance.
Full View into the End User’s Entire Transaction Journey
Specifically, the retailer needed an enterprise NetScaler ADC monitoring product that could scale to support packet throughput above 10,000 NetScaler transactions per second. And, a product their enterprise IT team could use to monitor the entire end user transaction journey — from website request initiation through transaction execution — and remediate transaction slowness or failures should they occur.
It’s about Business Too
To help people like product category managers, the retailer’s IT operations team uses the reports in Goliath for NetScaler to see how many requests are being made to each of the retailer’s product category VIPs (Virtual IP addresses), which show high latency or are generating error codes, which servers are generating errors such as timeouts, and how the traffic is distributed to the backend servers.
The retailer’s IT Operations team attempted to use NetScaler Insight Center to get the visibility and data they needed to effectively manage their NetScaler ADCs which typically can process over 10,000 eCommerce website transactions per second through the Citrix NetScaler Application Delivery Controller. Unfortunately, Insight Center was unable to scale to meet their enterprise IT needs, so they turned to Goliath for help.
In addition to being able to monitor the end user’s entire transaction journey (and not lose business to their competitor Amazon), the retailer needed a NetScaler performance and IT analytics solution that provided answers by giving them real-time dashboards, threshold based alerts, historical reports with unrestricted time frame reports, the ability to export performance data, and collaborate in real time.
The Retailer’s Environment Challenge: Managing NetScalers ADC & up to 10,000 NetScaler transactions per second
In the retailer’s environment there are 10 SDX NetScalers (see above), each load balancing multiple VIPs (Virtual IP addresses). Each VIP represents a product category in their retail operation (groceries, pharmacy, electronics, housewares, clothing, etc.).
Behind each VIP is one or more application servers. There are over 400 servers in total. Each cluster of app servers is connected to a database. All of this infrastructure controls information for over 10 million individual product items.
During normal operations, about 3,000 to 4,000 NetScaler requests are handled each second. During peak shopping hours that rate climbs to 10,000 NetScaler VIP transactions per second. Requests for product information come in from a variety of sources: the retailer’s eCommerce website, mobile apps, inventory management systems, etc.
The Retailer’s Solution
When this retail giant needed to get insight into their environment, they turned to Goliath for NetScaler to collect AppFlow messages broadcasted by their SDX NetScalers.
The retailer uses Goliath for NetScaler to understand the transaction behavior including performance, traffic routing from user to content loaded from the back end server, and reporting.
Each NetScaler request generates AppFlow messages containing the following data:
- The VIP
- The actual server selected
- The requested URL
- The returned status code
- The response in milliseconds
Goliath for NetScaler in turn, collects all of this information, stores it in a database, and provides a real-time dashboard showing the flow of traffic and details about latency, VIP activity, error codes, etc. At a glance, IT staff can see which servers are taking the longest to respond.
Management Gets Access to Critical Historical Performance Reporting
By using the historical reports feature in Goliath for NetScaler, the retailer’s management can get full visibility into their environment’s performance. Specifically, their management can now see:
- Requests being made on each VIP (this data is critical to the managers of the different product categories, like pharmacy, home products, etc.)
- Traffic distribution to the back-end web servers
- VIPs showing the highest latency
If requests coming to a VIP are taking too long, this indicates that additional app servers are needed on the VIP. If latency is low, the number of servers can be reduced:
- VIPs generating errors
An HTTP 404 error code would indicate that one of the upstream systems is requesting information for a product that doesn’t exist:
- Server errors
Errors such as Timeout on a server indicates server trouble or misconfiguration:
- Traffic pattern trending over time. By generating reports over long timelines, management can get a crystal clear picture of how many requests are being processed. This gives them insight into future expansion needs
The retailer is able to use Goliath for NetScaler to get this data without changing a single application server component, deploying agents, or causing negative impact on the actual server performance.
Why Goliath for NetScaler Solved the Retailer’s Problems
Because Goliath for NetScaler scales to support packet throughput above 10,000 NetScaler transactions per second while providing visibility into each step of the end user’s journey from website request initiation through transaction execution, the retailer’s IT operations team knew the technology would solve their problems.
There are several key points to derive from this success story:
- Goliath is the only product in the market, that can handle 10,000+ NetScaler transactions per second
- It has real-time threshold based alerts that are triggered by slow transaction performance
- Collecting all of the data is possible without changing a single application server component
- All of the performance data and metrics are available via AppFlow from the existing NetScalers
- Gathering the data has zero impact on the actual server performance
- No agents are needed to collect the data
By deploying Goliath’s technology, the retailer’s enterprise IT team can monitor the performance of their 10 SDX NetScalers, gain full visibility into the entire transaction journey in a single pane of glass, from the end user request through to the backend application server, and get alerted to interruptions or latency that would cause an end user to abandon their transaction.