The second paper we looked at in UU is Amazon’s 2007 paper onDynamo. Dynamo is an example of a new type of database dubbed NoSQL and Riak is an open-source implementation of the Dynamo architecture. Studying Dynamo is worthwhile for a number of reasons:

• It combines a lot of recent ideas in distributed systems. These ideas are worth learning in their own right to avoid mistakes likeReddit’s when building scalable systems.
• Since Riak is basically Dynamo, knowledge of Dynamo is directly applicable.
• Understanding the design trade-offs in Dynamo provides a way to understand the rest of the NoSQL space.

So, What is NoSQL?

In the old days everyone used relational databases and it was good. Then along came the web, and with the web a tidal wave of data, and things were not good. The tradeoffs made by relational databases (maintaining the famous ACID properties) made them unsuitable for tasks where response time and availability were paramount. This is the case for many web applications. For example, it doesn’t really matter if my Facebook status updates aren’t immediately visible to all my friends, but it does matter if my browser hangs for a minute while the back-end tries to get a write lock on the status table.

NoSQL databases make a different set of tradeoffs, and achieve different performance characteristics as a result. Typically, NoSQL databases focus on scalability, fast response times, and availability, and give up atomicity and consistency. This tradeoff is formalised via the CAP Theorem, which states that a distributed system cannot provide consistency, availability, and partition tolerance all at the same time (although two out of three of these properties are achievable at once). Dynamo provides availability and partition tolerance at the expense of consistency. Other NoSQL databases may make different tradeoffs. SQL databases typically provide consistency and availability at the expense of partition tolerance.

Reading the Paper

The Dynamo paper can be difficult to read. The main issue we had is that the authors don’t always motivate the different components of the system. For example, consistent hashing is one of the earlier concepts introduced in the paper, but it is difficult to see why it is used and how it contributes to increased availability until later on. It is best to approach each section of the article as a self-contained idea, and wait until the end to see how they are combined. It took us two sessions to get through the paper, so don’t be surprised if you find it slow going.

Setting Out the Shop

The paper starts by laying out the properties required of Dynamo. We’ve talked about the tradeoff between consistency, availability, and partition tolerance above. Some of the other properties are:

• Cost-effectiveness. This is important but often overlooked. You’ll sometimes see supporters of relational databases arguing that if people got some real database hardware they’d never need NoSQL. The problem with real hardware is it’s expensive. If my 20-CPU database server is at full capacity I have to drop another $20’000 just to handle another 5% increase in traffic. I probably can’t get next day delivery on this type of server, either. With a system like Dynamo I can just boot up another $500/yr virtual machine.
• Dynamo is a key-value store. This means that there are no foreign keys and hence no joins: the application must provide all of this, or more likely use a denormalised data representation. Furthermore, Dynamo sees its data as opaque binary blobs, so search is only possible using primary keys. Other NoSQL databases make different choices: MongoDB and CouchDB are document-oriented stores, meaning that data is stored as a JSON-like tree of keys and values; HBase and Cassandra store data as tuples, like a relational database, but without foreign keys.
• Low configuration, and fully distributed design. These two go hand-in-hand. A fully distributed design means all nodes are the same, and thus have the same configuration. It also means there is no single point of failure, another desirable feature. Again, different systems take different approaches. For example,MongoDB and most relational databases have a master/slave setup in which one machine has special “master” significance. Obviously in this setup different machines have different configurations.

Big Ideas

Dynamo is the fusion of a lot of ideas that are have developed in the field of distributed systems. Rather than duplicate the paper I want to discuss four points that I found interesting:

• Consistent hashing
• Dynamo’s implementation
• Amazon’s quality metric
• Feedback control for balancing tasks

Consistent Hashing

If you take one point from Dynamo, let it be the usefulness ofconsistent hashing. The basic idea of consistent hashing is to decouple the value of a key from the machine it is stored on. If you do this you can add and remove machines from your data store without breaking anything. If you don’t, you’re in a world of pain.

Consistent hashing is best explained via an example of doing it wrong. Say you have N machines serving as your data store. Given a key you want to work out which machine stores the data. A simple way to do so (which is what Reddit did) is to calculate key mod N. Now suppose due to increased load you want to add a machine in your data store. Now key mod (N+1) won’t give the same result, so you can’t find your data any more. To fix this you have to flush out the data and reinsert it, which will take a long time. Or you can use consistent hashing from the outset.

In consistent hashing you arrange the space of hash keys into a ring. Each server inserts a token into the ring, and is responsible for keys that lie in the range from it’s token to the nearest preceding token. This is illustrated in the image to the left. The small circles indicate the tokens, and the colours the segments of the hash ring allocated to each server.

Adding a new server only requires coordination with the server that previously occupied that part of the hash space. In the original consistent hashing paper tokens were inserted at random. For Dynamo it was found that a more structured system worked better. I’ll leave the details of this and other issues (in particular, routing and replication) to the paper.

Non-blocking IO

The section on Dynamo’s implementation will be interesting to PL geeks. If you’ve ever rolled your eyes at the manual continuation-passing style inflicted by Javascript then you might at least crack a wry smile when you read about essentially the same technique being used in Dynamo. There is an interesting debate to be had on the virtues of non-blocking IO vs thread-per-connection. At the moment my opinion is non-blocking IO is a necessary evil given kernels written in unsafe languages (and hence expensive context switches). Erlang does a good job of presenting a simple programming model with its light-weight threads, but achieving decent SMP performance can be hard due to the mismatch between application and OS threads. It’s my hope that languages like Rust will give a pragmatic solution to this dilemma.

Amazon’s Quality Metric

Although it isn’t part of the main thrust of the paper, I found it interesting that Amazon measure response time and other variables at the 99.9% percentile. Amazon have a very good reputation, and for other companies looking to achieve the same stature it is good to know the goal to aim for.

Feedback Control for Balancing Tasks

I’ve recently implemented feedback control (in particular, proportional error control) for a database connection pool. (I’ll blog about this in a bit.) It’s interesting that Dynamo uses a similar method to balance tasks within each node (Section 6.5). I think we’re going to see more self-regulating systems in the future. The work atRADLab is a good example of what might make it into production in a few years.

By scheduling tasks itself Dynamo is performing a task typically handled by the operating system. I think in the future this will be more commonplace, with the distinction between operating system and application program becoming increasingly blurred. TheManaged Runtime Initiative is one project that aims to do this.