I'm using SqlServer to drive a WPF application, I'm currently using NHibernate and pre-read all the data so it's cached for performance reasons. That works for a single client app, but I was wondering if there's an in memory database that I could use so I can share the information across multiple apps on the same machine. Ideally this would sit below my NHibernate stack, so my code wouldn't have to change. Effectively I'm looking to move my DB from it's traditional format on the server to be an in memory DB on the client.
Note I only need select functionality.
I would be incredibly surprised if you even need to load all your information in memory. I say this because, just as one example, I'm working on a Web app at the moment that (for various reasons) loads thousands of records on many pages. This is PHP + MySQL. And even so it can do it and render a page in well under 100ms.
Before you go down this route make sure that you have to. First make your database as performant as possible. Now obviously this includes things like having appropriate indexes and tuning your database but even though are putting the horse before the cart.
First and foremost you need to make sure you have a good relational data model: one that lends itself to performant queries. This is as much art as it is science.
Also, you may like NHibernate but ORMs are not always the best choice. There are some corner cases, for example, that hand-coded SQL will be vastly superior in.
Now assuming you have a good data model and assuming you've then optimized your indexes and database parameters and then you've properly configured NHibernate, then and only then should you consider storing data in memory if and only if performance is still an issue.
To put this in perspective, the only times I've needed to do this are on systems that need to perform millions of transactions per day.
One reason to avoid in-memory caching is because it adds a lot of complexity. You have to deal with issues like cache expiry, independent updates to the underlying data store, whether you use synchronous or asynchronous updates, how you give the client a consistent (if not up-to-date) view of your data, how you deal with failover and replication and so on. There is a huge complexity cost to be paid.
Assuming you've done all the above and you still need it, it sounds to me like what you need is a cache or grid solution. Here is an overview of Java grid/cluster solutions but many of them (eg Coherence, memcached) apply to .Net as well. Another choice for .Net is Velocity.
It needs to be pointed out and stressed that something like NHibernate is only consistent so long as nothing externally updates the database and that there is exactly one NHibernate-enabled process (barring clustered solutions). If two desktop apps on two different PCs are both updating the same database with NHibernate the caching simply won't work because the persistence units simply won't be aware of the changes the other is making.
http://www.db4o.com/ can be your friend!
Velocity is an out of process object caching server designed by Microsoft to do pretty much what you want although it's only in CTP form at the moment.
I believe there are also wrappers for memcached, which can also be used to cache objects.
You can use HANA, express edition. You can download it for free, it's in-memory, columnar and allows for further analytics capabilities such as text analytics, geospatial or predictive. You can also access with ODBC, JDBC, node.js hdb library, REST APIs among others.
I have a series of Oracle databases that need to access each other's data. The most efficient way to do this is to use database links - setting up a few database links I can get data from A to B with the minimum of fuss. The problem for me is that you end up with a tightly-coupled design and if one database goes down it can bring the coupled databases with it (or perhaps part of an application on those databases).
What alternative approaches have you tried for sharing data between Oracle databases?
Update after a couple of responses...
I wasn't thinking so much a replication, more on accessing "master data". For example, if I have a central database with currency conversion rates and I want to pull a rate into a separate database (application). For such a small dataset igor-db's suggestion of materialized views over DB links would work beautifully. However, when you are dynamically sampling from a very large dataset then the option of locally caching starts to become trickier. What options would you go for in these circumstances. I wondered about an XML service but tuinstoel (in a comment to le dorfier's reply) rightly questioned the overhead involved.
Summary of responses...
On the whole I think igor-db is closest, which is why I've accepted that answer, but I thought I'd add a little to bring out some of the other answers.
For my purposes, where I'm looking at data replication only, it looks like Oracle BASIC replication (as opposed to ADVANCED) replication is the one for me. Using materialized view logs on the master site and materialized views on the snapshot site looks like an excellent way forward.
Where this isn't an option, perhaps where the data volumes make full table replication an issue, then a messaging solution seems the most appropriate Oracle solution. Oracle Advanced Queueing seems the quickest and easiest way to set up a messaging solution.
The least preferable approach seems to be roll-your-own XML web services but only where the relative ease of Advanced Queueing isn't an option.
Streams is the Oracle replication technology.
You can use MVs over database links (so database 'A' has a materialized view of the data from database 'B'. If 'B' goes down, the MV can't be refreshed but the data is still in 'A').
Mileage may depend on DB volumes, change volumes...
It looks to me like it's by definition tightly coupled if you need simultaneous synchronous access to multiple databases.
If this is about transferring data, for instance, and it can be asynchronous, you can install a message queue between the two and have two processes, with one reading from the source and the other writing to the sink.
The OP has provided more information. He states that the dataset is very large. Well how large is large? And how often are the master tables changed?
With the use of materialized view logs Oracle will only propagate the changes made in the master table. A complete refresh of the data isn't necessary. Oracle streams also only communicate the modifications to the other side.
Buying storage is cheap, so why not local caching? Much cheaper than programming your own solutions.
An XML service doesn't help you when its database is not available so I don't understand why it would help? Oracle has many options for replication, explore them.
edit
I've build xml services. They provide interoperability between different systems with a clear interface (contract). You can build a xml service in C# and consume the service with Java. However xml services are not fast.
Why not use Advanced Queuing? Why roll your own XML service to move messages (DML) between Oracle instances - It's already there. You can have propagation move messages from one instance to another when they are both up. You can process them as needed in the destination servers. AQ is really rather simple to set up and use.
Why do they need to be separate databases?
Having a single database/instance with multiple schemas might be easier.
Keeping one database up (with appropriate standby databases etc) will be easier than keeping N up.
What kind of immediacy do you need and how much bi-directionality? If the data can be a little older and can be pulled from one "master source", create a series of simple ETL scripts run on a schedule to pull the data from the "source" database into the others.
You can then tailor the structure of the data to feed the needs of the client database(s) more precisely and you can change the structure of the source data until you're blue in the face.
I'm working on a web app that is somewhere between an email service and a social network. I feel it has the potential to grow really big in the future, so I'm concerned about scalability.
Instead of using one centralized MySQL/InnoDB database and then partitioning it when that time comes, I've decided to create a separate SQLite database for each active user: one active user per 'shard'.
That way backing up the database would be as easy as copying each user's small database file to a remote location once a day.
Scaling up will be as easy as adding extra hard disks to store the new files.
When the app grows beyond a single server I can link the servers together at the filesystem level using GlusterFS and run the app unchanged, or rig up a simple SQLite proxy system that will allow each server to manipulate sqlite files in adjacent servers.
Concurrency issues will be minimal because each HTTP request will only touch one or two database files at a time, out of thousands, and SQLite only blocks on reads anyway.
I'm betting that this approach will allow my app to scale gracefully and support lots of cool and unique features. Am I betting wrong? Am I missing anything?
UPDATE I decided to go with a less extreme solution, which is working fine so far. I'm using a fixed number of shards - 256 sqlite databases, to be precise. Each user is assigned and bound to a random shard by a simple hash function.
Most features of my app require access to just one or two shards per request, but there is one in particular that requires the execution of a simple query on 10 to 100 different shards out of 256, depending on the user. Tests indicate it would take about 0.02 seconds, or less, if all the data is cached in RAM. I think I can live with that!
UPDATE 2.0 I ported the app to MySQL/InnoDB and was able to get about the same performance for regular requests, but for that one request that requires shard walking, innodb is 4-5 times faster. For this reason, and other reason, I'm dropping this architecture, but I hope someone somewhere finds a use for it...thanks.
The place where this will fail is if you have to do what's called "shard walking" - which is finding out all the data across a bunch of different users. That particular kind of "query" will have to be done programmatically, asking each of the SQLite databases in turn - and will very likely be the slowest aspect of your site. It's a common issue in any system where data has been "sharded" into separate databases.
If all the of the data is self-contained to the user, then this should scale pretty well - the key to making this an effective design is to know how the data is likely going to be used and if data from one person will be interacting with data from another (in your context).
You may also need to watch out for file system resources - SQLite is great, awesome, fast, etc - but you do get some caching and writing benefits when using a "standard database" (i.e. MySQL, PostgreSQL, etc) because of how they're designed. In your proposed design, you'll be missing out on some of that.
Sounds to me like a maintenance nightmare. What happens when the schema changes on all those DBs?
http://freshmeat.net/projects/sphivedb
SPHiveDB is a server for sqlite database. It use JSON-RPC over HTTP to expose a network interface to use SQLite database. It supports combining multiple SQLite databases into one file. It also supports the use of multiple files. It is designed for the extreme sharding schema -- one SQLite database per user.
One possible problem is that having one database for each user will use disk space and RAM very inefficiently, and as the user base grows the benefit of using a light and fast database engine will be lost completely.
A possible solution to this problem is to create "minishards" consisting of maybe 1024 SQLite databases housing up to 100 users each. This will be more efficient than the DB per user approach, because data is packed more efficiently. And lighter than the Innodb database server approach, because we're using Sqlite.
Concurrency will also be pretty good, but queries will be less elegant (shard_id yuckiness). What do you think?
If you're creating a separate database for each user, it sounds like you're not setting up relationships... so why use a relational database at all?
If your data is this easy to shard, why not just use a standard database engine, and if you scale large enough that the DB becomes the bottleneck, shard the database, with different users in different instances? The effect is the same, but you're not using scores of tiny little databases.
In reality, you probably have at least some shared data that doesn't belong to any single user, and you probably frequently need to access data for more than one user. This will cause problems with either system, though.
I am considering this same architecture as I basically wanted to use the server side SQLLIte databases as backup and synching copy for clients. My idea for querying across all the data is to use Sphinx for full-text search and run Hadoop jobs from flat dumps of all the data to Scribe and then expose the results as webservies. This post gives me some pause for thought however, so I hope people will continue to respond with their opinion.
Having one database per user would make it really easy to restore individual users data of course, but as #John said, schema changes would require some work.
Not enough to make it hard, but enough to make it non-trivial.
Locked. This question and its answers are locked because the question is off-topic but has historical significance. It is not currently accepting new answers or interactions.
So I'm using an app that stores images heavily in the DB. What's your outlook on this? I'm more of a type to store the location in the filesystem, than store it directly in the DB.
What do you think are the pros/cons?
I'm in charge of some applications that manage many TB of images. We've found that storing file paths in the database to be best.
There are a couple of issues:
database storage is usually more expensive than file system storage
you can super-accelerate file system access with standard off the shelf products
for example, many web servers use the operating system's sendfile() system call to asynchronously send a file directly from the file system to the network interface. Images stored in a database don't benefit from this optimization.
things like web servers, etc, need no special coding or processing to access images in the file system
databases win out where transactional integrity between the image and metadata are important.
it is more complex to manage integrity between db metadata and file system data
it is difficult (within the context of a web application) to guarantee data has been flushed to disk on the filesystem
As with most issues, it's not as simple as it sounds. There are cases where it would make sense to store the images in the database.
You are storing images that are
changing dynamically, say invoices and you wanted
to get an invoice as it was on 1 Jan
2007?
The government wants you to maintain 6 years of history
Images stored in the database do not require a different backup strategy. Images stored on filesystem do
It is easier to control access to the images if they are in a database. Idle admins can access any folder on disk. It takes a really determined admin to go snooping in a database to extract the images
On the other hand there are problems associated
Require additional code to extract
and stream the images
Latency may be
slower than direct file access
Heavier load on the database server
File store. Facebook engineers had a great talk about it. One take away was to know the practical limit of files in a directory.
Needle in a Haystack: Efficient Storage of Billions of Photos
This might be a bit of a long shot, but if you're using (or planning on using) SQL Server 2008 I'd recommend having a look at the new FileStream data type.
FileStream solves most of the problems around storing the files in the DB:
The Blobs are actually stored as files in a folder.
The Blobs can be accessed using either a database connection or over the filesystem.
Backups are integrated.
Migration "just works".
However SQL's "Transparent Data Encryption" does not encrypt FileStream objects, so if that is a consideration, you may be better off just storing them as varbinary.
From the MSDN Article:
Transact-SQL statements can insert, update, query, search, and back up FILESTREAM data. Win32 file system interfaces provide streaming access to the data.
FILESTREAM uses the NT system cache for caching file data. This helps reduce any effect that FILESTREAM data might have on Database Engine performance. The SQL Server buffer pool is not used; therefore, this memory is available for query processing.
File paths in the DB is definitely the way to go - I've heard story after story from customers with TB of images that it became a nightmare trying to store any significant amount of images in a DB - the performance hit alone is too much.
In my experience, sometimes the simplest solution is to name the images according to the primary key. So it's easy to find the image that belongs to a particular record, and vice versa. But at the same time you're not storing anything about the image in the database.
The trick here is to not become a zealot.
One thing to note here is that no one in the pro file system camp has listed a particular file system. Does this mean that everything from FAT16 to ZFS handily beats every database?
No.
The truth is that many databases beat many files systems, even when we're only talking about raw speed.
The correct course of action is to make the right decision for your precise scenario, and to do that, you'll need some numbers and some use case estimates.
In places where you MUST guarantee referential integrity and ACID compliance, storing images in the database is required.
You cannot transactionaly guarantee that the image and the meta-data about that image stored in the database refer to the same file. In other words, it is impossible to guarantee that the file on the filesystem is only ever altered at the same time and in the same transaction as the metadata.
As others have said SQL 2008 comes with a Filestream type that allows you to store a filename or identifier as a pointer in the db and automatically stores the image on your filesystem which is a great scenario.
If you're on an older database, then I'd say that if you're storing it as blob data, then you're really not going to get anything out of the database in the way of searching features, so it's probably best to store an address on a filesystem, and store the image that way.
That way you also save space on your filesystem, as you are only going to save the exact amount of space, or even compacted space on the filesystem.
Also, you could decide to save with some structure or elements that allow you to browse the raw images in your filesystem without any db hits, or transfer the files in bulk to another system, hard drive, S3 or another scenario - updating the location in your program, but keep the structure, again without much of a hit trying to bring the images out of your db when trying to increase storage.
Probably, it would also allow you to throw some caching element, based on commonly hit image urls into your web engine/program, so you're saving yourself there as well.
Small static images (not more than a couple of megs) that are not frequently edited, should be stored in the database. This method has several benefits including easier portability (images are transferred with the database), easier backup/restore (images are backed up with the database) and better scalability (a file system folder with thousands of little thumbnail files sounds like a scalability nightmare to me).
Serving up images from a database is easy, just implement an http handler that serves the byte array returned from the DB server as a binary stream.
Here's an interesting white paper on the topic.
To BLOB or Not To BLOB: Large Object Storage in a Database or a Filesystem
The answer is "It depends." Certainly it would depend upon the database server and its approach to blob storage. It also depends on the type of data being stored in blobs, as well as how that data is to be accessed.
Smaller sized files can be efficiently stored and delivered using the database as the storage mechanism. Larger files would probably be best stored using the file system, especially if they will be modified/updated often. (blob fragmentation becomes an issue in regards to performance.)
Here's an additional point to keep in mind. One of the reasons supporting the use of a database to store the blobs is ACID compliance. However, the approach that the testers used in the white paper, (Bulk Logged option of SQL Server,) which doubled SQL Server throughput, effectively changed the 'D' in ACID to a 'd,' as the blob data was not logged with the initial writes for the transaction. Therefore, if full ACID compliance is an important requirement for your system, halve the SQL Server throughput figures for database writes when comparing file I/O to database blob I/O.
One thing that I haven't seen anyone mention yet but is definitely worth noting is that there are issues associated with storing large amounts of images in most filesystems too. For example if you take the approach mentioned above and name each image file after the primary key, on most filesystems you will run into issues if you try to put all of the images in one big directory once you reach a very large number of images (e.g. in the hundreds of thousands or millions).
Once common solution to this is to hash them out into a balanced tree of subdirectories.
Something nobody has mentioned is that the DB guarantees atomic actions, transactional integrity and deals with concurrency. Even referentially integrity is out of the window with a filesystem - so how do you know your file names are really still correct?
If you have your images in a file-system and someone is reading the file as you're writing a new version or even deleting the file - what happens?
We use blobs because they're easier to manage (backup, replication, transfer) too. They work well for us.
The problem with storing only filepaths to images in a database is that the database's integrity can no longer be forced.
If the actual image pointed to by the filepath becomes unavailable, the database unwittingly has an integrity error.
Given that the images are the actual data being sought after, and that they can be managed easier (the images won't suddenly disappear) in one integrated database rather than having to interface with some kind of filesystem (if the filesystem is independently accessed, the images MIGHT suddenly "disappear"), I'd go for storing them directly as a BLOB or such.
At a company where I used to work we stored 155 million images in an Oracle 8i (then 9i) database. 7.5TB worth.
Normally, I'm storngly against taking the most expensive and hardest to scale part of your infrastructure (the database) and putting all load into it. On the other hand: It greatly simplifies backup strategy, especially when you have multiple web servers and need to somehow keep the data synchronized.
Like most other things, It depends on the expected size and Budget.
We have implemented a document imaging system that stores all it's images in SQL2005 blob fields. There are several hundred GB at the moment and we are seeing excellent response times and little or no performance degradation. In addition, fr regulatory compliance, we have a middleware layer that archives newly posted documents to an optical jukebox system which exposes them as a standard NTFS file system.
We've been very pleased with the results, particularly with respect to:
Ease of Replication and Backup
Ability to easily implement a document versioning system
If this is web-based application then there could be advantages to storing the images on a third-party storage delivery network, such as Amazon's S3 or the Nirvanix platform.
Assumption: Application is web enabled/web based
I'm surprised no one has really mentioned this ... delegate it out to others who are specialists -> use a 3rd party image/file hosting provider.
Store your files on a paid online service like
Amazon S3
Moso Cloud Storage
Another StackOverflow threads talking about this here.
This thread explains why you should use a 3rd party hosting provider.
It's so worth it. They store it efficiently. No bandwith getting uploaded from your servers to client requests, etc.
If you're not on SQL Server 2008 and you have some solid reasons for putting specific image files in the database, then you could take the "both" approach and use the file system as a temporary cache and use the database as the master repository.
For example, your business logic can check if an image file exists on disc before serving it up, retrieving from the database when necessary. This buys you the capability of multiple web servers and fewer sync issues.
I'm not sure how much of a "real world" example this is, but I currently have an application out there that stores details for a trading card game, including the images for the cards. Granted the record count for the database is only 2851 records to date, but given the fact that certain cards have are released multiple times and have alternate artwork, it was actually more efficient sizewise to scan the "primary square" of the artwork and then dynamically generate the border and miscellaneous effects for the card when requested.
The original creator of this image library created a data access class that renders the image based on the request, and it does it quite fast for viewing and individual card.
This also eases deployment/updates when new cards are released, instead of zipping up an entire folder of images and sending those down the pipe and ensuring the proper folder structure is created, I simply update the database and have the user download it again. This currently sizes up to 56MB, which isn't great, but I'm working on an incremental update feature for future releases. In addition, there is a "no images" version of the application that allows those over dial-up to get the application without the download delay.
This solution has worked great to date since the application itself is targeted as a single instance on the desktop. There is a web site where all of this data is archived for online access, but I would in no way use the same solution for this. I agree the file access would be preferable because it would scale better to the frequency and volume of requests being made for the images.
Hopefully this isn't too much babble, but I saw the topic and wanted to provide some my insights from a relatively successful small/medium scale application.
SQL Server 2008 offers a solution that has the best of both worlds : The filestream data type.
Manage it like a regular table and have the performance of the file system.
It depends on the number of images you are going to store and also their sizes. I have used databases to store images in the past and my experience has been fairly good.
IMO, Pros of using database to store images are,
A. You don't need FS structure to hold your images
B. Database indexes perform better than FS trees when more number of items are to be stored
C. Smartly tuned database perform good job at caching the query results
D. Backups are simple. It also works well if you have replication set up and content is delivered from a server near to user. In such cases, explicit synchronization is not required.
If your images are going to be small (say < 64k) and the storage engine of your db supports inline (in record) BLOBs, it improves performance further as no indirection is required (Locality of reference is achieved).
Storing images may be a bad idea when you are dealing with small number of huge sized images. Another problem with storing images in db is that, metadata like creation, modification dates must handled by your application.
I have recently created a PHP/MySQL app which stores PDFs/Word files in a MySQL table (as big as 40MB per file so far).
Pros:
Uploaded files are replicated to backup server along with everything else, no separate backup strategy is needed (peace of mind).
Setting up the web server is slightly simpler because I don't need to have an uploads/ folder and tell all my applications where it is.
I get to use transactions for edits to improve data integrity - I don't have to worry about orphaned and missing files
Cons:
mysqldump now takes a looooong time because there is 500MB of file data in one of the tables.
Overall not very memory/cpu efficient when compared to filesystem
I'd call my implementation a success, it takes care of backup requirements and simplifies the layout of the project. The performance is fine for the 20-30 people who use the app.
Im my experience I had to manage both situations: images stored in database and images on the file system with path stored in db.
The first solution, images in database, is somewhat "cleaner" as your data access layer will have to deal only with database objects; but this is good only when you have to deal with low numbers.
Obviously database access performance when you deal with binary large objects is degrading, and the database dimensions will grow a lot, causing again performance loss... and normally database space is much more expensive than file system space.
On the other hand having large binary objects stored in file system will cause you to have backup plans that have to consider both database and file system, and this can be an issue for some systems.
Another reason to go for file system is when you have to share your images data (or sounds, video, whatever) with third party access: in this days I'm developing a web app that uses images that have to be accessed from "outside" my web farm in such a way that a database access to retrieve binary data is simply impossible. So sometimes there are also design considerations that will drive you to a choice.
Consider also, when making this choice, if you have to deal with permission and authentication when accessing binary objects: these requisites normally can be solved in an easier way when data are stored in db.
I once worked on an image processing application. We stored the uploaded images in a directory that was something like /images/[today's date]/[id number]. But we also extracted the metadata (exif data) from the images and stored that in the database, along with a timestamp and such.
In a previous project i stored images on the filesystem, and that caused a lot of headaches with backups, replication, and the filesystem getting out of sync with the database.
In my latest project i'm storing images in the database, and caching them on the filesystem, and it works really well. I've had no problems so far.
Second the recommendation on file paths. I've worked on a couple of projects that needed to manage large-ish asset collections, and any attempts to store things directly in the DB resulted in pain and frustration long-term.
The only real "pro" I can think of regarding storing them in the DB is the potential for easy of individual image assets. If there are no file paths to use, and all images are streamed straight out of the DB, there's no danger of a user finding files they shouldn't have access to.
That seems like it would be better solved with an intermediary script pulling data from a web-inaccessible file store, though. So the DB storage isn't REALLY necessary.
The word on the street is that unless you are a database vendor trying to prove that your database can do it (like, let's say Microsoft boasting about Terraserver storing a bajillion images in SQL Server) it's not a very good idea. When the alternative - storing images on file servers and paths in the database is so much easier, why bother? Blob fields are kind of like the off-road capabilities of SUVs - most people don't use them, those who do usually get in trouble, and then there are those who do, but only for the fun of it.
Storing an image in the database still means that the image data ends up somewhere in the file system but obscured so that you cannot access it directly.
+ves:
database integrity
its easy to manage since you don't have to worry about keeping the filesystem in sync when an image is added or deleted
-ves:
performance penalty -- a database lookup is usually slower that a filesystem lookup
you cannot edit the image directly (crop, resize)
Both methods are common and practiced. Have a look at the advantages and disadvantages. Either way, you'll have to think about how to overcome the disadvantages. Storing in database usually means tweaking database parameters and implement some kind of caching. Using filesystem requires you to find some way of keeping filesystem+database in sync.