The Case for Faster Search

Improving Search Results with Index-Based Search
Sphinx Logo
When we say Heroic Support, we mean it – and it doesn’t stop with our support staff. The entire company is always working to constantly improve how we help our customers. That motivation is what led us to update the search function on one of our internal sites, HelpDesk. HelpDesk is the program that our technicians use to communicate with customers via support tickets. We knew that a more efficient search function for HelpDesk would mean faster problem solving for our customers.

The Problem – Inefficient Search

Our HelpDesk program uses a very large database full of all our past and current tickets from customers. When customers have a problem with which they need help, a ticket is usually created to log the problem and how it was resolved. Over 15 years in business means we have a lot of tickets. Not to mention, HelpDesk was set up over a decade ago with a now-outdated search function – one that used sequence scans. Sequence scans, if you’re not aware, require your data to be searched, row by row, every time a query is made. Of course, the more data you have, the longer this takes, and our HelpDesk database is very large. Sometimes a simple search would take minutes to load – a time that may not seem like very long until you’re on the phone with a customer who desperately needs help solving an issue. We knew that our outdated search function was preventing our support staff from helping our customers as quickly and efficiently as possible.

The Search for Something New

With all of our searchable content already in a database, we were ready to begin looking for a new search tool. We knew that sequence scans were a very inefficient way to use a database for search – especially if you have a large database. We were looking for search tools that go through your database only once and create an index of information. After that index is created, queries are typically answered in milliseconds, rather than in the minutes or hours that can be expected with sequence scanning. This type of search can mean the difference between hours of lost time and a search function that is actually beneficial to users.

There are a lot of excellent options for a site search program, but they all differ in overall performance. Choosing the right tool for our HelpDesk meant looking carefully at our own needs and measuring them against each tool. We considered PostGreSQL internal full text search, Apache Solr, Elasticsearch and Sphinx and focused on their ease of use, scalability, flexibility and performance. Of course, the primary feature of all of these search tools is how they index your data, eliminating the slow and painful process of scanning your database sequentially.

The Solution – Sphinx’s Index Search

In the end, we chose Sphinx, a program that is used by many top websites internationally, including Craigslist. A vast improvement over our old system, we chose Sphinx because it fit our requirements perfectly. Not only did it speed up our data retrieval enormously, but setup was also easy and it worked with our current system, which allowed us to avoid deploying new libraries or writing new code. Sphinx offered us high performance, incredible speed, reliability, ease of use, and easy scalability. The switch to Sphinx means a better performing HelpDesk for Liquid Web employees, allowing them to provide the exceptional Heroic Support® for which we are known.

No matter where your search function will be used, on an internal site or a public-facing one, efficient search is vital to success. The purpose of these search tools is to make your content more accessible to users, allowing them to skip over traditional navigation and jump straight to the information they need. Especially if your search function is on a public-facing site, a slow result could be devastating. A customer who doesn’t find the information they’re searching for within a few seconds is a lost customer.

At Liquid Web, losing customers isn’t an option – and one way we keep our customers happy is to always improve our customer service. Updating the search function for HelpDesk was a logical improvement for our team and a great way to improve our already Heroic Customer Service. No matter what industry you’re in, customer service is key. Happy customers mean a successful business.

If you’d like to learn more about how Liquid Web strives to keep our customers happy through Heroic Support and quality web hosting, Live Chat or call our team at 1-800-580-4985.