问题
And what kind of alternative strategies do you use for avoiding LazyLoadExceptions?
I do understand that open session in view has issues with:
- Layered applications running in different jvm\'s
- Transactions are committed only at the end, and most probably you would like the results before.
But, if you know that your application is running on a single vm, why not ease your pain by using an open session in view strategy?
回答1:
Because sending possibly uninitialised Proxies, especially collections, in the view layer and triggering hibernate loading from there can be troubling from both a performance and understanding point of view.
Understanding:
Using OSIV 'pollutes' the view layer with concerns related to the data access layer.
The view layer is not prepare to handle a HibernateException
which may happen when lazy loading, but presumably the data access layer is.
Performance:
OSIV tends to tug proper entity loading under the carpet - you tend not to notice that your collections or entities are lazily initialised ( perhaps N+1 ). More convenience, less control.
Update: see The OpenSessionInView antipattern for a larger discussion regarding this subject. The author lists three important points:
- each lazy initialization will get you a query meaning each entity will need N + 1 queries, where N is the number of lazy associations. If your screen presents tabular data, reading Hibernate’s log is a big hint that you do not do as you should
- this completely defeats layered architecture, since you sully your nails with DB in the presentation layer. This is a conceptual con, so I could live with it but there is a corollary
- last but not least, if an exception occurs while fetching the session, it will occur during the writing of the page: you cannot present a clean error page to the user and the only thing you can do is write an error message in the body
回答2:
For a longer description, you can read my Open Session In View Anti-Pattern article. Otherwise, here's a summary for why you shouldn't use Open Session In View.
Open Session In View takes a bad approach to fetching data. Instead of letting the business layer decide how it’s best to fetch all the associations that are needed by the View layer, it forces the Persistence Context to stay open so that the View layer can trigger the Proxy initialization.
- The
OpenSessionInViewFilter
calls theopenSession
method of the underlyingSessionFactory
and obtains a newSession
. - The
Session
is bound to the TransactionSynchronizationManager. - The
OpenSessionInViewFilter
calls thedoFilter
of thejavax.servlet.FilterChain
object reference and the request is further processed - The DispatcherServlet is called, and it routes the HTTP request to the underlying
PostController
. - The
PostController
calls thePostService
to get a list ofPost
entities. - The
PostService
opens a new transaction, and theHibernateTransactionManager
reuses the sameSession
that was opened by theOpenSessionInViewFilter
. - The
PostDAO
fetches the list ofPost
entities without initializing any lazy association. - The
PostService
commits the underlying transaction, but theSession
is not closed because it was opened externally. - The
DispatcherServlet
starts rendering the UI, which, in turn, navigates the lazy associations and triggers their initialization. - The
OpenSessionInViewFilter
can close theSession
, and the underlying database connection is released as well.
At a first glance, this might not look like a terrible thing to do, but, once you view it from a database perspective, a series of flaws start to become more obvious.
The service layer opens and closes a database transaction, but afterward, there is no explicit transaction going on. For this reason, every additional statement issued from the UI rendering phase is executed in auto-commit mode. Auto-commit puts pressure on the database server because each statement must flush the transaction log to disk, therefore causing a lot of I/O traffic on the database side. One optimization would be to mark the Connection
as read-only which would allow the database server to avoid writing to the transaction log.
There is no separation of concerns anymore because statements are generated both by the service layer and by the UI rendering process. Writing integration tests that assert the number of statements being generated requires going through all layers (web, service, DAO), while having the application deployed on a web container. Even when using an in-memory database (e.g. HSQLDB) and a lightweight web server (e.g. Jetty), these integration tests are going to be slower to execute than if layers were separated and the back-end integration tests used the database, while the front-end integration tests were mocking the service layer altogether.
The UI layer is limited to navigating associations which can, in turn, trigger N+1 query problems. Although Hibernate offers @BatchSize for fetching associations in batches, and FetchMode.SUBSELECT to cope with this scenario, the annotations are affecting the default fetch plan, so they get applied to every business use case. For this reason, a data access layer query is much more suitable because it can be tailored for the current use case data fetch requirements.
Last but not least, the database connection could be held throughout the UI rendering phase(depending on your connection release mode) which increases connection lease time and limits the overall transaction throughput due to congestion on the database connection pool. The more the connection is held, the more other concurrent requests are going to wait to get a connection from the pool.
So, either you get the connection held for too long, either you acquire/release multiple connections for a single HTTP request, therefore putting pressure on the underlying connection pool and limiting scalability.
Spring Boot
Unfortunately, Open Session in View is enabled by default in Spring Boot.
So, make sure that in the application.properties
configuration file, you have the following entry:
spring.jpa.open-in-view=false
This will disable OSIV, so that you can handle the LazyInitializationException the right way.
回答3:
transactions can be committed in the service layer - transactions are not related to OSIV. It's the
Session
that stays open, not a transaction - running.if your application layers are spread across multiple machines, then you pretty much can't use OSIV - you have to initialize everything you need before sending the object over the wire.
OSIV is a nice and transparent (i.e. - none of your code is aware that it happens) way to make use of the performance benefits of lazy loading
回答4:
I wouldn't say that Open Session In View is considered a bad practice; what gives you that impression?
Open-Session-In-View is a simple approach to handling sessions with Hibernate. Because it's simple, it's sometimes simplistic. If you need fine-grained control over your transactions, such as having multiple transactions in a request, Open-Session-In-View is not always a good approach.
As others have pointed out, there are some trade-offs to OSIV -- you're much more prone to the N+1 problem because you're less likely to realize what transactions you're kicking off. At the same time, it means you don't need to change your service layer to adapt to minor changes in your view.
回答5:
If you're using an Inversion of Control (IoC) container such as Spring, you may want to read up on bean scoping. Essentially, I'm telling Spring to give me a Hibernate Session
object whose life cycle spans the entire request (i.e., it gets created and destroyed at the start and end of the HTTP request). I don't have to worry about LazyLoadException
s nor closing the session since the IoC container manages that for me.
As mentioned, you will have to think about N+1 SELECT performance issues. You can always configure your Hibernate entity afterwards to do eager join loading in places where performance is an issue.
The bean scoping solution is not a Spring-specific. I know PicoContainer offers the same capability and I'm sure other mature IoC containers offer something similar.
回答6:
In my own experience, OSIV is not so bad. The only arrangement I made is using two different transactions: - the first, opened in "service layer", where I have the "business logic" - the second opened just before the view rendering
回答7:
I just did a post on some guidelines as to when to use open session in view in my blog. Check it out if your interested.
http://heapdump.wordpress.com/2010/04/04/should-i-use-open-session-in-view/
回答8:
I am v. rusty on Hibernate.. but I think its possible to have multiple transactions in one Hibernate session. So your transaction boundaries do not have to be the same as session start/stop events.
OSIV, imo, primarily is useful because we can avoid writing code for starting a 'persistence context' (a.k.a. session) every time the request needs to make a DB access.
In your service layer, you will probably need to make calls to methods which have different transaction needs, such as 'Required, New Required, etc.' The only thing these methods need is that somebody (i.e the OSIV filter) has started up the persistence context, so that only thing they have to worry about is - "hey give me the hibernate session for this thread.. I need to do some DB stuff".
回答9:
This won't help too much but you can check my topic here: * Hibernate Cache1 OutOfMemory with OpenSessionInView
I have some OutOfMemory issues because of OpenSessionInView and a lot of entities loaded, because they stay in Hibernate cache level1 and are not garbage collected (i load a lot of entities with 500 items per page, but all entities stay in cache)
来源:https://stackoverflow.com/questions/1103363/why-is-hibernate-open-session-in-view-considered-a-bad-practice