Several of Allcloud’s projects this year have focused on enterprise search and integration in the cloud.
When choosing a back-end search platform, the general requirements are common: the ability to ingest a lot of content, query its full text quickly with flexible options, and secure the data in transit and at rest.
Sometimes you plan to write all of the crawling and indexing logic yourself. Other times, you may prefer a pre-built connector to a storage tier. The amount of data in your search index, the granularity of that data, and your management requirements may vary wildly.
To help make decisions between cloud search platforms, I’ve organized common decision criteria into a simple table:
|
Factor
(Last Assessed)
|
Azure Search
|
AWS CloudSearch
|
AWS Elasticsearch
|
Elasticsearch (Self-Hosted)
|
|
Homepage
|
Azure Search
|
AWS CloudSearch
|
AWS Elasticsearch
|
Elasticsearch
|
|
Underlying search platform
|
Elasticsearch
|
Solr
|
Elasticsearch
|
Elasticsearch
|
|
Disposition
|
Platform-as-a-Service (PaaS)
|
Platform-as-a-Service (PaaS)
|
Managed Infrastructure-as-a-Service (IaaS)
|
Infrastructure-as-a-Service (IaaS)
|
|
Built-in Data Source Connectors
|
Azure SQL, Azure Cosmos DB, Azure BLOB Storage
|
AWS DynamoDB, AWS S3
|
AWS CloudWatch, AWS DynamoDB, AWS Kinesis, AWS Kinesis Firehose, AWS S3
|
Many, with plugins
|
|
Built-in File Formats Supported
|
CSV, EML, Microsoft Office (DOC/DOCX, XLS/XLSX, PPT/PPTX, MSG), HTML, JSON, PDF, RTF, TXT, XML, ZIP
|
CSV, JSON, XML
|
JSON
|
Many, with plugins
|
|
Languages Supported
|
56
|
35
|
33
|
33
|
|
Plugins Supported
|
No
|
No
|
Yes, some
|
Yes, many
|
|
Application Programming Interfaces
|
REST APIs and many SDKs
|
REST APIs and many SDKs
|
REST APIs and many SDKs
|
REST APIs and many SDKs
|
|
Maximum Number of Instances
|
12 (36 with replicas)
|
50
|
20
|
No limit
|
|
Maximum Number of Documents
|
1.4 Billion
|
Unknown (Up to 10,000 batches / day)
|
2 Billion / index
|
2 Billion / index
|
|
Maximum Number of Fields
|
1,000
|
200
|
1,000
|
1,000
|
|
Maximum Document Size
|
16 MB
|
1 MB
|
2 GB
|
2 GB
|
|
Maximum Bulk Operation Size
|
1,000 documents / 16 MB
|
1,000 documents / 5 MB
|
100 MB
|
Configurable
|
|
Nexted Object Fields Support
|
No
|
No
|
Yes
|
Yes
|
|
TLS Encryption in Transit
|
Yes
|
Yes
|
Yes
|
Requires X-Pack add-on
|
|
Encryption at Rest
|
Unclear
|
No
|
Yes
|
Possible with third-party software
|
|
Automatic Scaling
|
No
|
Yes
|
No
|
No
|
|
Pricing Model
|
Managed nodes (per hour)
|
EC2 infrastructure instances (Per hour)
|
EC2 Infrastructure Instances (per hour)
|
Core software is free; X-Pack requires a license; cloud fees are up to you
|
|
Price Range Per Month, Per Instance (US)
|
Free - $1,999.87
|
$43.90 - $559.49
|
$13.39 - $4,670.83
|
Variable
|
