JSR107 (Java Caching API) Update – Lots Happening

I have been very busy the last few months getting JSR107 fired up.

Just to remind you JSR 107 is the Java Temporary Caching API. It is designed to be vendor neutral and will allow for easy change over of implementations in much the same way as JPA or JDBC. In this way it will allow the community to choose the best open source or commercial implementation that best meets their business requirements.

Both Terracotta and Oracle have committed resources to getting this done. Right now that is myself and Yannis Cosmadopoulos from Oracle. Things are moving fast so I thought I would give a status update on where we are at.

JSR107 Early Draft Working Specification Available for Viewing

Work has been going steadily on the spec.

We are drafting this spec in the open on Google Docs. While a work in progress, it is now around 40 pages in length.

See https://docs.google.com/document/d/1YZ-lrH6nW871Vd9Z34Og_EqbX_kxxJi55UrSn4yL2Ak/edit?hl=en

Please keep checking back as this is changing on a daily basis.

We welcome ideas and feedback. Please join the JSR107 public mailing list to do so.

Summary of Scope

The specification covers the following areas:

Object Cache The API will cache objects. Classes that implement Serializable may be stored outside the JVM and potentially shared among JVMs connected by a network but Serializable will not be required.

Format independence The API will specify keys and values but will not limit their types.

Implementation independence The underlying implementation is hidden from API users. An SPI is used to configure a Cache Provider.

Support for Flexible Implementations Though the specification does not require any implementation and a simple in-process implementation will meet the specification, issues raised by distributed caching and storage in Serialized form outside the heap will be dealt with so that those implementations with those features will work well.

Java SE The specification will work with Java SE.

Java EE The specification will work within Java EE. This specification is targeted at inclusion in Java EE 7.

Generics The specification will make use of generic interfaces.

Annotations The specification will define runtime cache annotations.

Declarative Cache Configuration Specifying the behaviour of the CacheManager or Caches in a non-programmatic way. This may take the form of a minimal lowest common denominator with vendor specific further configuration, or it might take the form of a variety of mechansims to inject vendor configuration.

Transactions Support for transactions, both local and XA will be defined but left as optional for implementers

There are many applied areas of caching. This specification will not deal with them. For example:

• Database Caching. JPA[7] deals with that.

• Servlet Caching

• Caching as a REST service


Active Membership

I am pleased to report that we have a very healthy membership. We reconfirmed the membership, added new members with expertise and also added as observers the leads from JSR342 Java EE7.

The following are active expert group members:

  • Greg Luck, Terracotta and co-spec lead
  • Cameron Purdy, Oracle and co-spec lead
  • Nikita Ivanov, Grid Gain

• Manik Surtani, Red Hat Middleware LLC

• Yannis Cosmadopolous, Oracle and working on the spec for Oracle

• Chris Berry

• Andy Piper (formerly BEA Systems rep)

• Jon Stevens

The following have been voted in they are moving through the JCP process:

• Eric Dalquist

• Ben Cotton, Citigroup

• David Mossakowski, Citigroup

The following are members of JSR342 and are observers of JSR107 (they get the expert group mails):

• Linda Demichiel, Oracle

• Roberto Chinnici, Oracle

GitHub Repositories

Similar to the specification we are also coding up the API in public with repositories on GitHub.


See https://github.com/jsr107/jsr107spec


Ehcache via the ehcache-jcache module has provided an implementation of the draft spec to where it got to, for the past 3 years. We have now moved it to GitHub and will develop it along with the spec. If anyone wants to help out please send me your GitHub id.

See https://github.com/jsr107/ehcache-jcache

Public Mailing List

If you want to stay in touch with JSR107, please join our Google Groups public mailing list:  http://groups.google.com/group/jsr107

The address is: jsr107@googlegroups.com

We copy most expert group emails to this list.


Comparative Technical Use Cases for Distributed Caches and NoSQL

I have been doing some NoSQL research lately. The first fruit of that work was a guest post on myNoSQL, Ehcache: Distributed Cache or NoSQL Store, which crisply distinguished between a Distributed Cache and NoSQL Stores.

In this article I am going to delve into the suitability of each for various technical use cases. I use the word “technical” because a usual use case is a business use case. Here we are interested in a set of features that allow a certain usage. In a follow up I hope to create a second, more business use case oriented table.

I welcome feedback on this, particularly from those with production experience.

Technical Use Case Distributed Cache NoSQL Key Value NoSQL Columnar NoSQL Graph NoSQL Document
Database Offload Excellent Poor (1) Poor (1) Poor (1) Poor (1)
Database Replacement Poor (2) Poor (3) Poor (3) Poor (3) Poor (3)
Weak Consistency Caching Excellent Average (2) Average Poor Poor
Eventual Consistency Caching Excellent (4) Average (5) Average (5) Average (5) Average (5)
Strongly Consistent Caching Excellent Poor Poor Poor Poor
ACID Transactional Caching Excellent Poor Poor Poor Poor
Low Latency Data Access Excellent Average (5) Average (5) Average (5) Average (5)
Big Data (6) Poor Excellent Excellent Excellent Excellent
Big Memory (7) Excellent (8) Poor Poor Poor Poor


  1. To offload the database you need to work in places and ways in which the database works. So for example you need to support transactions if they are being used and you need a place to plug in to avoid a ton of work like Hibernate or OpenJPA. NoSQL stores don’t do that.
  2. Distributed caches may not provide long term persistence and management of data. They are also often limited in size so may not be able to store all of the data.
  3. It is not clear that NoSQL is a full database replacement. The “Not Just SQL” as an alternative expansion of the acronym, something widely accepted by the NoSQL community, acknowledges this. The lack of SQL, the lack of ACID, sophisticated operations tools and so on, mean that NoSQL itself is not great at being a replacement. Rather, if you can rethink your need for a database to needing persistence, and you can change your application code, then it comes into play.
  4. In a node to the elegant CAP trade off allowed by eventual consistency, Ehcache 2.4.1, due out the end of March adds this consistency mode.
  5. Distributed Caches store hot data in process. You might think of memcache as a distributed cache, which it claims to be but it does not store data in -process – it is always over the network. And NoSQL is always over the network. In most R + W > N strategies, R is greater than one, so that multiple network reads are required and the caller must wait for R reads where each read is to a different server which will have a varying response. Distributed Ehcache has latencies of < 1 ms whereas the average for NoSQL is 5-10ms. This is also why NoSQL gets an average for Weak Consistency Caching. A cache should be fast.
  6. “Big Data” is a moving target that is today generally understood to start at a few dozen terabytes and go up into petabytes. The current implementation of Ehcache has been used to manage datasets up to 2 TB which is just at the starting point of Big Data. The whole point of NoSQL is Big Data, so they get full marks in this area.
  7. “Big Memory” is also a moving target and is early on it’s use as a term. We define it to mean using the physical limits of the hardware. For many architectures this has not been possible. With Java the issue was first 32 bits and then now the limitation is garbage collection. We overcame that issue with our BigMemory architecture, using storage in off-heap byte buffers in September 2010.
  8. Caches tends to be memory-resident. BigMemory allows in-memory densities per physical server up to their limits, which is 2TB for the current generation of commodity hardware from Dell, HP and Oracle but much lower due to their architectures which require full CPU population to achieve maximum memory. Although not all vendors are similarly constrained: Cisco UCS boxes allow more memory per CPU, so that for example they can do 384GB with 2 CPUs. NoSQL stores focus on persistency and have small in-memory server side caches. They focus on speeding up disk reads and writes by for example doing append only.

Holy Moly, Batman. Apple upgraded my Maven

I ran a Maven build this morning and it broke. Strange, as I had not changed anything. Maven’s ability to simply break because of a change to a non-versioned dependency or even more arcane, a versioned dependency with a non-versioned dependency of it’s own (a transitive dependency) is legendary.  So I thought that was the trouble. On my other machine I ran a build and the Maven output “looked” different. Stranger still. So I did a mvn version on both.

Whoa! My Big Mac was 2.11 and my MacBook Pro was 3.0.2.

It came down in the “Java for Mac  OS X  10.6 Update 4” which was released on 10 March.

Now a lot of folks are not ready to move to 3. For myself firstly I don’t need the grief until it is rock solid. Secondly, I use the site plugin and reporting extensively and this is all getting a big makeover in 3 which is not yet done.

There are lots of ways to go back to your old version.

My was is to add the old version into the front of my PATH in .bash_profile:


News on JSR107 (JCACHE) and JSR342 (Java EE 7)


JSR342 was created on 14 March 2011. JSR107, or JCACHE, is included: In JSR342’s words:

The following new JSRs will be candidates for inclusion in the Java EE 7 platform:

Concurrency Utilities for Java EE (JSR-236)
JCache (JSR-107)

Isn’t JSR107 inactive?

But how could this happen if JSR107 is inactive?

Well the answer is that we are reactivating it. Oracle (various staff) and Terracotta (mostly me) have started work on the specification with the hope of having a draft spec for review by 20 April. To actually make sure it happens Oracle have allocated resources to work on this project. They are being led by Cameron Purdy, who is co-spec lead along with myself of JSR107 and the founder of Coherence. And of course I founded and still continue to lead Ehcache as CTO of it at Terracotta.

To be officially reactivated we need to submit the draft spec. So reactivation should happen on 20 April.

Motivations for finishing JSR107

Today there are two leading widely scoped frameworks for developing enterprise applications: Spring and Java EE. With the release of Spring 3.1, Spring, heavily influenced by Ehcache Annotations for Spring, has significantly enhanced their caching API. It is easier for Spring because they are a single vendor framework and can do things outside of a standards process. Java EE is still lacking any general purpose caching API. There are some use-specific APIs scattered throughout such as in JPA, but nothing developers can write to. And I know there is a significant need for a general purpose API. So, Java EE 7 wants a general purpose caching API, and this is the primary reason for finishing JSR107.

Another reason is that in-process caching is now heavily commoditised but not standardised. There are more than 20 open source in-process caching projects and another 5 or so commercial distributed caches. But if a user wants to change implementations, they need to change their code. This is akin to database access not having the JDBC standard. So we need to provide a standard API so that users can change caching implementations at low cost.

Scope of JSR107

There has been a bit of discussion about this, but it is most likely that the scope will be as it has been plus two new areas:

  1. Generics – similar to collections, allow caches to be created with defined keys and values
  2. Annotations and integration with JSR342 – allow caching annotations so that for example the return value from any functional method can be cached

The draft specification as it has existed for a few years is available under a net.sf.jsr107cache package name on SourceForge. And Ehcache provides an implementation of that draft spec via ehcache-jcache.

Get Involved

It seems a good time to look at the expert group and make a general invitation for new members.

If you are interested and would like to spend some time on this, please email me at gluck At gregluck.com and I can explain how to start. Additions to membership of the JSR107 are by voting of the existing members.

Ehcache: Distributed Cache or NoSQL Store?

Is Ehcache a NoSQL store? No, I would not characterise it as that, but I have seen it used for some NoSQL use cases. In these situations it compared very well — with higher performance and more flexible consistency than the well-known NoSQL stores. Let me explain.

Read more

Ehcache 2.4 with the new Search feature is out!

Ehcache 2.4 launches today. The big new feature right in the core of Ehcache 2.4 is Search.

It uses a new fluent API which looks like this:

Results results = cache.createQuery().includeKeys().addCriteria(age.eq(32).and(gender.eq(“male”))).execute();

In short, it lets you further offload the database. With Ehcache now supporting up to 2TB and linear scale-out you can do more than ever.

What is searchable?

You can search against predefined indexes of keys, values, or attributes extracted from values.

Attributes can be extracted using JavaBeans conventions or by specifying a method to call.

For example to declare a cache searchable and extract age as a JavaBean and gender as a method out of a Person class:

Caches can also be made searchable programmatically. And Custom Value Extractors can be created so that you can index and search against virtually anything you put in the cache.

Search Query Language

Ehcache Search introduces EQL, a fluent, Object Oriented query language which we call EQL, following DSL principles, which should feel familiar and natural to Java programmers.

Here is a full example. Search for men whose names start with “Greg”, and then order the results by age. Don’t return more than 10 results. We want to include keys and values in the results. Finally we iterate through the Results.

Query query = cache.createQuery();
query.addCriteria(name.ilike(“Greg*”).and(gender.eq(Gender.MALE))).addOrderBy(age, Direction.ASCENDING).maxResults(10);

Results results = query.execute();
System.out.println(” Size: ” + results.size());
for (Result result : results.all()) {
System.out.println(“Got: Key[” + result.getKey()
+ “] Value class [” + result.getValue().getClass()
+ “] Value [” + result.getValue() + “]”);

EQL is very rich. There is a large number of  Criteria such as ilike, lt, gt, between and or which you use to build up complex queries. There are also Aggregators such as min, max, average, sum and count which will summarise the results.

Like NoSQL, EQL executes against a single cache – there are no joins. If you need to combine the results of searches from two caches, you can perform two searches  and then combine the results yourself.

Standalone and Distributed

Search is built into Ehcache core. It works with standalone in-process caching and will work for distributed caches in the forthcoming Terracotta 3.5 platform release which goes GA in March and is available as a release candidate now.

Distributed cache search is indexed and executes on the Terracotta Server Array using a scatter gather pattern. The EQL is sent to each cache partition (the scatter), returning partial results (the gather) to the requesting Ehcache node which then combines the results and presents them to the caller. Terracotta servers utilise precomputed indexes to speed queries.

Indeed, the distributed cache performance has an important property: searches execute in O(logN)/partitions time. So if you have 50GB of cache in one partition which takes 40ms to search and then you double the data, you can hold the execute time constant by simply adding another partition. Generally, you can hold execution time constant for any size of data with this approach.

The standalone cache takes a different approach. Most in-process caches are relatively small. And Ehcache is lightning fast. We don’t use indexes but instead visit each element in the cache a maximum on once, resolving the EQL. It takes 5ms to run a typical query against a 10,000 element cache. Generally the performance is O(N) but even a 1 million entry cache will take less than a second to search using this approach.

Sample Use Cases

Caches can also be made searchable programmatically. And Custom Value Extractors can be created so that you can index and search against virtually anything you put in the cache.

Database Search Offload

Take a shipping company that creates 50GB of consignment notes per week. Customers search by consignment note id but also by addressee name. Most searches (95%) are done within two weeks of the creation of a consignment note.   The consignment notes get stored in a relational database that grows and grows. Searches against the database now take 650ms which take enquiry outside it’s SLA.

Solution: Put the last two weeks of data in the cache. Index by consignment note id, first name, last name and date.  Search the cache first and only search the database if there the consignment note is not found. This takes about  50ms and provides a 95% database offload.

Real-Time Analytics Search

In-house analytics engines processes large amounts of data and compute some result from it. The results need to be queried very quickly to enable processing of a within a business transaction. And the results need to be updated through the day in response to business events. Some examples are credit card fraud scoring, or a holding position in a trading application.

Create a distributed cache and index it as required. Various roll-ups are cached and updated after the system of record has been written to with new transactions. Use Ehcache’s bulk loading mode to quickly upload the results of overnight analytics runs.

Searches execute much more quickly than it would take to compute positions from scratch using the system of record, enabling the real-time analytics.

More Information

You can get Ehcache 2.4 from ehcache.org. Search is fully documented along with downloadable demos and performance tests here.

Something new under the Sun: A new way of doing a cache invalidation protocol with Oracle 11g

Just when you think the database will never get smarter, it does. And the database that matters most to Enterprise Developers is Oracle.

What happens to coherency between your distributed cache and the database when another application changes the database? If it is a Java application, you can add Ehcache to it and configure it to connect to the distributed cache. But this might require work you do not want to do right now. Or there might be tens of apps involved. Or there might be scripts which are run by DBAs from time to time. Or it could be another language. We have the Cache Server, where you can expose a distributed cache via REST or SOAP and then invalidation becomes as simple as sending a HTTP Delete to the Element resource. See the Cache Server Documentation for more on this.

For a few years, MySQL users have had access to cache invalidation for Memcached using libmemcached. The database calls back to memcached. Brian Aker mentioned to me he was adding this into MySQL a few years ago. See http://www.mysqlconf.com/mysql2009/public/schedule/detail/6277 for a decent introduction.

For Oracle, back in 2005 I tried to use their message queue integration to achieve the same effect. Back then it didn’t even work. I have heard that it works now, but it is a very messy solution with lots of moving parts.

Fortunately, starting from 11g Release 1 (11.1), the Oracle JDBC driver provides support for the Database Change Notification feature of Oracle Database. Using this functionality of the JDBC drivers, multitier systems can take advantage of the Database Change Notification feature to maintain a data cache as up-to-date as possible, by receiving invalidation events from the JDBC drivers. See  the Database Change Notification chapter in the Oracle docs for details.

This lets you achieve a new and very simple way of doing a cache invalidation protocol to keep your cache and the database in sync.

You create a DatabaseChangeListener and register it with your connection:

I like this so much I think we might add a standard DatabaseChangeListener to Ehcache for it.

Ehcache: The Year In Review

Is it that time of year again. A time for reflection and of New Year’s resolutions. I therefore thought this was good time to reflect on what has been happening with Ehcache.

Ehcache and Terracotta got together in August 2009. We got our first combined release done with Ehcache backed by Terracotta three months later. That was really only the beginning. We have done 5 major releases to date with of course another one in the cooker right now.

The big news is that Ehcache and Terracotta together have been a great success. Most Terracotta users now use it as a distributed backing store for Ehcache. Most of Terracotta’s revenue comes from that use case. This means that Ehcache’s needs are driving the evolution of Terracotta.

At the same time, Ehcache as an open source project has seen a huge investment. Most of the features added are designed to work in Ehcache open source as well as with Terracotta as the backing distributed store.

So let’s look at what we added in 2010 and what is in the cooker so far for 2011.

What we did in 2010

New Hibernate Provider

When we merged Terracotta had just done a Hibernate provider and Ehcache had one too. Neither supported the new Hibernate 3.4 SPI which uses CacheRegionFactories instead of CachingProviders. So we combined the two into a new implementation and added support for the new SPI at the same time. This meant that for the first time Ehcache supported all of the cache strategies, including transactional, in Hibernate and importantly supported them across a caching cluster.  (http://ehcache.org/documentation/hibernate.html)

XA Transactions

Right now XA transactions have fallen from favour partly because of flaky support for them in XAResource implementations out there. But what if you could create a canonically correct implementation that could be absolutely relied on in the world’s most demanding transactional applications? We hired one of the Java world’s foremost experts in transactions (Ludovic Orban, the author of the Bitronix transaction manager) and came out with just that. We challenge anyone to prove that our implementation is not correct and does not fully deal with all failure scenarios. If you need to be absolutely sure that the cache is in sync with your other XAResources with a READ_COMMITTED isolation level you have come to the right place. (http://ehcache.org/documentation/jta.html)

Terabyte Scale Out

Ehcache backed by Terracotta initially held keys in memory in each application that had Ehcache in it. This effectively limited the size of the caching clusters that could be created. With a new storage strategy, we blew the lid of that and stopped storing the keys in memory. The result – horizontal scaling to the terabyte level.

Write-through, behind and every which way

What happens when you have off-loaded reads from your database but now your writes are killing you? The answer is write-behind. You write to the cache. It calls a CacheWriter which you implement and connect to the cache which is called periodically with batches of transactions. In your CacheWriter you open a transaction write the batch and then close the transaction. Much easier for the database. And all done in HA with the write-behind queue is safe because it is stored on the Terracotta cluster.

More caching in more places

We were really happy to extend our support for popular frameworks. During the year:

  • Ehcache became the default caching provider for Grails
  • We created an OpenJPA provider
  • We created Ruby Gems for JRuby and Rails 2 and 3 caching providers
  • We created a Google App Engine module

Acknowledgement of the CAP Theorem

Originally Ehcache with Terracotta was consistent above all else. During the year we flexed both ways to allow CAP tradeoffs to be made by our users. We added XA and manual locking modes on the even stricter side and we added an unlocked reads view of a cache and even coherent=false for scale out without coherence on the looser side. And you can choose this cache by cache. There is a tradeoff between latency and consistency, so you choose the highest speed you can afford for a particular cache.

And rather than just blocking on a network partition, we added NonStopCache, a cache decorator which allows you to choose whether to favour availability or consistency.


BigMemory was a big hit. It surprised a lot of people and frankly it surprised us. We were looking to solve our own GC issues in the Terracotta server and found something that was more generally useful than that one use case. So we added BigMemory to Ehcache standalone as well as the server. In the server we have lifted our field engineering recommendation from 20GB of storage per server partition to 100Gb. And we have tested BigMemory itself out to 350GB and it works great!

A new Disk Store

Let’s say you are using a 100GB in Ehcache standalone. When you restart your JVM you want the cache to be there otherwise it might take hours or days to repopulate such a large cache. So we created a new DiskStore that keeps up with BigMemory. It writes at the rate of 10MB/s. So when it is time to shutdown your JVM it just needs to do a final sync and then your are done. And it starts up straight away and gradually loads data into memory. A nice complement to BigMemory and very important.

Ehcache Monitor/Terracotta Dev Console Improvements

For those using Ehcache standalone we have only ever had a JMX API. That is fine but we found many people built their own web app to gather stats. So we did the same and the result was Ehcache Monitor. One of the highlights is the charts including a chart of estimated memory use per cache.

The Terracotta Developer Console got an Ehcache panel, and as we added features to Ehcache we added more to the panel. If you are using Ehcache with a backing Terracotta store then it is a full featured tool which gives you deep introspection.

What is Coming in 2011

Speed, speed and more speed

What does everybody want? More speed. We are splitting hairs in our concurrency model to enable as much speed as possible for each use case. We now have two and will be adding more modes to allow the best tuning for each use case.


Ehcache is based on a Map API. Maps have keys and values. They have a peculiar property – you need to know the key to access the value. What if you want to search for a key, or you want to index values in numerous ways and search those. All of this is coming to Ehcache in February 2011 and is available right now in beta. Oh and one cool thing: search performance is O(log n/partitions). So as your data grows and spreads out onto more Terracotta server partitions, your search performance stays constant! (http://ehcache.org/documentation/search.html)

New Transaction Modes

We already did the hard one: XA. Now we are adding Local Transactions. If you just want transactionality within the caches in your CacheManager and there are no other XAResources, you can use a Local Transaction. It will be three times faster than an XA cache. (http://ehcache.org/documentation/jta.html)

.NET Client

Quite a few customers use Java but also some .NET. And they want to be able to share caches. We have lots of users happily using ehcache for cross-platform use cases, but are planning on extending our cross-platform support still further – for example with a native .NET client

Bigger BigMemory

We are looking at ongoing speedups and testing against larger and larger memory sizes for BigMemory. We are also looking to provide further speed in BigMemory by allowing pluggable Serialization strategies. This will allow our users to use their Serialization framework of choice – and there are now quite a few.