Recent AWS Customer Success Stories & Videos

More AWS Customer Success Stories...

« AWS Documentation Now Available on the Kindle | Main | Microsoft SharePoint Server on AWS Reference Architecture White Paper »


Feed You can follow this conversation by subscribing to the comment feed for this post.


Can you index the contents of PDF files/.doc files stored on S3?


Does it support S3? do you plan to index S3 and offer the search for it?


How about different languages? Atm its not working for russian.

Anton Babenko

My impression is that this is a good service, but it won't be enough for medium and large web applications, because it lacks some good to have features like spacial queries and support for normalization rules in different languages. This is exactly what I need for my app, which I run on AWS stack already and looking towards utilizing Elastic Search.

Is there a plan to add these features in the near future?

Mark Kubacki

In order to completely outsource the searching-service and query the AWS servers by JavaScript without using proxy servers, you need to add some more features:
- either an optional function call, which encloses the JSON response (e.g.: ...&callback=displayResults)
- or CNAMEs for search domains

I did browse through the documentation but have not found that, yet.

A Facebook User

Do you provide tools to easily integrate the search results into an existing web site according to its theme? It seems that this step of presenting the search results in a user-friendly way is also important to really decrease the cost of searching. For example, Acquia provides a Drupal module on tops of its cloud search that facilitates theming the search results.

It seems that your solution gives a bit more of flexibility concerning stopwords, stemming and synonyms. Do you think it can work well for another language than English? If yes, how much time do you think it can be adapted to another language than English?


Jeff --

thanks for walkthru. We've been waiting for this to happen eagerly. One thing that we miss, however, is a pre-existing Web index that we can get access to. It seems that all you can search is 8 M of your own documents. Can we sing up as early customers for Web search as a service prior to its release?

Sincerely, Linda

Gaurav DCosta

Congrats! Awesome Timing guys! We started evaluating SphinxSearch for StrikeBase [our SaaS product] a week ago. We now plan to be using CloudSearch extensively. Could I have a date by which CloudSearch will be added to the AWS PHP SDK[it doesnt seem to be there right now]?


Aric Rosenbaum

Very interesting. Good stuff. Is there support for languages other than English? How about multi-byte languages such as Chinese and Arabic?

Olivier Janssens

You guys never fail to amaze me :)

Do you have geospatial search coming up? Right now we have over 20 dedicated sphinx instances running just for this purpose. It would be incredibly nice if you could deliver a solution for geospatial search!


A Facebook User

Looks like more fun than building and managing my own Solr cluster, Jeff! Hope you're good. :-)

Jeff Greenhouse

Sounds like a great offering, but it looks like the only option for getting data into it is to feed up all of the documents in their completed form. Is there any option for this service to crawl a website or URL? If not, that would be a logical thing to add.

While many CMS systems come with a built-in site search, website owners often want to present blended results that combine site pages (and blog posts, etc.) with more structured results. Being able to get all that search power in one place would be an even bigger value.

Just my two cents.

Jeff Greenhouse
President, 201 Proof -


Do you have a comparison/checklist chart against Solr & ElasticSearch?

Rafael Costa

Sounds amazing! What about autocomplete searches? Does it support autocomplete?



This is really great. All I would say is that I have a few clients who I have built custom Apache Solr-based web applications for, and who host those on virtual cloud servers that cost not that much more than the Small rate you have here just for search. Whilst I can definitely see the added value Amazon brings with scaling and I would love to move my Apache Solr clients to this, it's cost prohibitive I think, my clients would not pay the extra right now.

Also, do you support custom boosting of certain document types for example, or boosting by field? We have 4 types of document and we boost one of them as more important than the others. Then, within that, we boost further on a field of that document type if it exists/set to a certain value.

Keen to learn more about the advanced features (custom query handler, how much can you configure query slop, proximity etc..) and whether you index NGRAMs?


I've been working on trying to implement Xapian for search for the last few months, this service is a godsend (just wish I knew about it before we wasted all of our time trying to roll it out ourselves!)

Sian Kit Tjie

Will the service support spatial search?


It would be good if you also displayed the AWS responses to peoples comments.


Can/will this be multi-tenant? What if I have hundreds or thousands of customer-specific data sets to index, but they can only be searched by the customer?


Is there any way to know ahead of time how large of a search instance will be needed and potentially stop the search before it ends up costing more than I expect ?


This is definitely a good service offering; however, the devil is in the details. I'll be playing with the service for the next couple of months to learn more about it to see if it's a possible fit for us. It would be nice if spatial was part of the mix, boosting, custom query handlers etc.

Congratulations Jeff. We were excited to use this service in beta and it was a great time saver over building our own search infrastructure with Solr / Lucine. Happy it is now launched and we can discuss publicly!

Matt Wood

Great comments - thanks everyone. Keep them coming.

If you have specific queries or feedback about CloudSearch, the best place to get them infront of the CloudSearch team is on the forum:

We have operators standing by.

Petro Sasnyk

Could you clarify. Are there any limitatitions to the quantity of search domains?
How are "search instance" and "search domain" related? Are these notions equal?
Currently, we are having a separate Lucene index for each registered user, users number is unlimited.
To migrate our current architecture, should we create dedicated search domain for each user, or domain quantity are limited and we should separate users date through filtering search queries and results?


Is this a better choice than using DynamoDB with custom Index (tables) do you think ? I'm developing an website with Elastic Beanstalk (Java) and Dynamodb rightnow - some of the tables have around 500,000 entries in there currently

Tahseen Ur Rehman Fida

It is not clear any where how many documents an instance can index. I know it depends on the size of the document some sort of calculator will be helpful. I am working on an application that needs to index millions of documents every month. So it will be great to know when I will hit max 10 instance limit.


If you add geospatial I am *in*, currently using SOLR and worried about my cluster.


Definitely faster and easier than setting up and managing a Solr instance. We were rapidly able to leverage this technology to index 10,000+ gene expression data sets in our application, more discussion at


How to create custom field type amazon cloud search like date, textgen, textTight, float

Jeff Barr

The CloudSearch forum at is the best place to ask technical questions!

The comments to this entry are closed.

Featured Events

The AWS Report

Brought to You By

Jeff Barr (@jeffbarr):

Jinesh Varia (@jinman):

Email Subscription

Enter your email address:

Delivered by FeedBurner

October 2016

Sun Mon Tue Wed Thu Fri Sat
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25 26 27 28 29
30 31