FILESTREAM – What and When

If you have to store files (BLOB) in MS SQL Server database, you might consider using FILESTREAM (started from MS SQL Server 2008). This is essentially enabling FILESTREAM attribute of varbinary (max) type column. Some other setups are necessary prior to that.

FILESTREAM combines MS SQL Server with NTFS file system. Database is good at reading/writing/processing small pieces (rows and columns) of information scattered around in the database files while file system is good at reading/writing files as a whole. File system start performing better when the files are larger than 1 MB.

By combining the two you can get the best of the both worlds, like getting past the low max size limit of database BLOB and use max NTFS size, good streaming performance of file system, integration with structured data (transactional consistency), better fragmentation management, manageability (backups etc can be done together) etc.

However if you are doing frequent small updates FILESTREAM performs poor. This is because here it has to work more – guarantee transactional consistency as well write to file system that essentially creates/writes a new copy and delete the old one. It also takes more time to take backup.

FILESTREAM data can be accessed in two ways: standard Transact-SQL and by using Win32 APIs. If you want the best performance of File System you need to use Win32 APIs. There are provisions for the Win32 APIs to work within the context of database transaction.

FILESTREAM BLOBS use file system cache and does not use database buffer pool. This improves query processing and overall database performance.

However do note that certain features like Transparent Data Encryption (TDE) are not supported with FILESTREAM.

To summarize, if you are to save files 1 MB or larger in database that will be read a lot with few small updates and that a middle tier is available where you can use Win32 APIs for accessing them then FILESTREAM is a good choice.

Reference

Next

Index

Author: Gopal Das

I am working as a Principal Consultant at JLT Interactive Pte Ltd, Singapore. I have a BS in Computer Science & Engineering from Khulna University and ME in Internet Science & Engineering from Indian Institute of Science (IISc). I have a few publications on Query Optimization in RDBMS in ACM SIGMOD, IEEE ICDE etc. I was a founding team member and VP Engineering of iTwin, a spinoff from A*STAR. I am working as a software engineer for 14 years. I am interested in Algorithms, Database, Security, Machine Learning, and Blockchain among others. I am a father of two children and live in Singapore. www.linkedin.com/in/dasgopal https://github.com/gopalcdas

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s