Apache Solr integration for nopCommerce - More than 900% improvement in average response time

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.
11 years ago
Hello,

We've developed an extension for nopCommerce that integrate Apache Solr with nopCommerce.

Apache Solr is a blazing fast opensource search engine that is employed by many of the world's largest site to power their Search Results as well as Catalog Navigation Pages.

The result is obviously better search results and faster site response time.

We've bring the same benefit for nopCommerce platform by integrating Apache Solr. This makes your nopCommerce based store faster, highly scalable & reliable. Apache Solr can be scaled easily both horizontally as well as vertically, so you can quickly scale your stores using this plugin.

Apart from that it fixes performance issue of nopCommerce, as all the catalog pages will be served from Solr index instead of database. This makes your site respond very very quickly, at the same time makes you a lot of room to grow your e-commerce store.

We request you to checkout nopAccelerate to find how it can help you accelerate your nopCommerce store and make it more profitable.

The extension is going to be available as three different products, aiming different needs of the different store. You can check more details by visiting our website.


Update

Getting Started Support Documentation published.

Update August 2013

1. nopAccelerate now supports nopCommerce 2.8, 3.0 and 3.1 including Multi-Store, Multi-Lingual, Multi-Currency features. Check all the features.

2. nopAccelerate updated it demo store to have 80,000+ dummy products in its demo store

3. nopAccelerate supports Ajax Filtering with Advanced Filtering UI, Price Range Filter, Multi-Select Facets, etc. and loads page without full refresh instantly. Demo of filtering features.

4. nopAccelerate offers drill-down search on Search Results page. You can offer filters using Product Specifications, Category, Manufacturer or Product Attribute filter. Check at the demo store.

5. nopAccelerate now supports spell check and offers an alternative suggestions if wrongly spelled term is entered by user. This is also popularly known as "Did you mean?". Check demo.

6. It now supports searching for the inflectional form of the searched term. Inflection is the modification of a word to express different grammatical categories such as tense, mood, voice, aspect, person, number, gender and case. So treating them as one. ex: treating stabilizing, stabilizer, stabilized as stabilize.

7. nopAccelerate now offers advanced language analysis to improve search results, which supports more than 30 languages

8. nopAccelerate now offers subscription plan for hosting Solr, so you now don't need to install and maintain Solr Server. Select subscription plan while ordering, install plugin to nopCommerce store, and configure it and you're ready to go! Note that your Solr Core can be hosted on highly optimized, secured, cloud Solr servers.

9. If you just want to improve your Search experience on your store, then you can start using nopAccelerate Search which is available at just USD 9 per month (includes a Solr Core). Contact Us to Order.


Moreover, nopAccelerate does not just optimizes Search, but also Category, Manufacturer and Product Tags Pages to make your overall store respond instantly even with a huge catalog.

If you are facing an issue scaling your store, contact us to see how we can help you.



We're open to your ideas, suggestions & reviews to make it even better. So feel free to comment here or on our website.

Thanks,
Krunal

nopAccelerate
11 years ago
Update

1. Features Comparison page is now added at: http://www.nopaccelerate.com/features-comparison/
2. Demo is now ready for you with 10,000 products in its catalog which can be accessed at: http://demo.nopaccelerate.com/
11 years ago
jariwalakrunal wrote:
Update

1. Features Comparison page is now added at: http://www.nopaccelerate.com/features-comparison/
2. Demo is now ready for you with 10,000 products in its catalog which can be accessed at: http://demo.nopaccelerate.com/

Very fast. Good job!!
11 years ago
The demo is hosted on very basic VPS with just 1 CPU & 2 GB RAM. And we have 10,000 products in catalog.

Moreover, we did a load testing involving numbers of request and tried to create a load of 10 concurrent requests per second. With nopAccelerate, we achieved a apx. 900% improvement in response time. Without nopAccelerate it was having 20,100 ms average response time, after installing nopAccelerate with same test and load, we got the average response time to 1940ms. Everything else was the same. That's a huge improvement without investing in hardware upgrade.

And we found almost 300% improvement in average response time for general browsing without any such load.

Average response time is the application response time measured using the New Relic. And we created a load using JMeter, for about an hour for both tests.

We'll be publishing more details very soon. Stay tuned.

Update

- We now have 83000 products in a demo store catalog.
11 years ago
Is the suggestions engine using a different search engine than the full search?

If I type "rin" into search box I get a bunch of suggestions about wRINkle free shirts, yet when I search for "rin"

http://demo.nopaccelerate.com/search?q=rin

I get nothing..

Is this a Solr limitation or some implementation bug?

Filip
11 years ago
fkierzek wrote:
Is the suggestions engine using a different search engine than the full search?

If I type "rin" into search box I get a bunch of suggestions about wRINkle free shirts, yet when I search for "rin"

http://demo.nopaccelerate.com/search?q=rin

I get nothing..

Is this a Solr limitation or some implementation bug?

Filip


Thank you Filip for having a look at that and providing us your valuable review.

Right now, the suggestions are what nopCommerce provide by default. Search Suggestions from Solr is something we'll be working very soon. I have sent you one site URL in PM where you can see it in action.

For Search results, it is using Solr's engine. Solr search is configurable and it is configured to search word and not the characters in the word, so if you search for rin, it will only show results that have rin as a word. Its a design decision that we took to offer better search results, but it is configurable and can be changed as you can define in Solr what fields to search and how to search.

Have a look at dozens of Solr Analyzers, Tokenizers & Token Filters http://wiki.apache.org/solr/AnalyzersTokenizersTokenFilters which can be used to further fine tune search results and offer better search-ability to your site visitors which is not possible with the full text search offered by MSSQL.

You look at any major e-commerce website and you'll easily find out that none of them is actually using full text search offered by MYSQL/MSSQL, and most of them are using search engine like Solr, Elastic Search, Lucene, Fast Search (Microsoft) or Google Search for eCommerce (Google Commerce Search). Solr is the most popular among them and available as Open Source and has capability to scale easily compared to other alternatives. If you look at prices of Fast Search or Google Commerce Search and compare their features with Solr then, I'm sure you'll love what Solr offers as a free search engine.. that's the reason its the most popular not only for e-commerce sites but all kinds of sites.

Most major e-commerce site's catalog and especially its faceted/layerd navigation is also powered by Solr.

Solr is very capable, and offers you XML based configuration where you can instruct Solr for processing search queries. So its completely under your control and you can decide how you want to use it.

Moreover, with this plugin we're using Solr as a database to speedup nopCommerce. Solr stores data as a structured document (think it as a database with single flat table), so you can imagine why its blazing fast and performant that respond queries well with in a 1-2 millisecond even for a very busy sites in production environment.

For one of the busiest site where Solr serving appx. 2400 calls per minute for nopCommerce site, we noticed an average response time of 0.178 ms (Yes that's more than 5600 times faster than 1ms). Have a look at it here. CPM is calls per minute.. that is actually queries performed into Solr.

I welcome your comments & review on what you think about it, and what features you want to see to be implemented to make it even better.

Regards
11 years ago
eadameg wrote:


Very fast. Good job!!


Thank you for your review.
11 years ago
Hey it is pretty fast !!
good job..

i got one suggestion..
i tried search in the demo store.

it doesn't provide the 'did you mean this' feature like google.


else its great !!
thumbs up...


Have a good day !!
11 years ago
supriya_salunkhe wrote:
Hey it is pretty fast !!
good job..

i got one suggestion..
i tried search in the demo store.

it doesn't provide the 'did you mean this' feature like google.


else its great !!
thumbs up...


Have a good day !!


Thank you Supriya for your review & complements.

It's called the Fuzzy Search. This is right now not developed. But it's a good feature to add. We've added it to our list of features that we'll be working in future. Solr already have this feature, we need to do some configuration in Solr and then few changes into the plugin to support this.

Once again, thank you for your reviews & suggestions. And Sign up to nopAccelerate. We'll be providing you the Trial & detailed information very soon.
11 years ago
jariwalakrunal wrote:


Thank you Supriya for your review & complements.

It's called the Fuzzy Search. This is right now not developed. But it's a good feature to add. We've added it to our list of features that we'll be working in future. Solr already have this feature, we need to do some configuration in Solr and then few changes into the plugin to support this.

Once again, thank you for your reviews & suggestions. And Sign up to nopAccelerate. We'll be providing you the Trial & detailed information very soon.


That would be great !!
I did sign up..
Looking forward for the trial version...

Have a good day !
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.