Transaction Semantics in Kudu

Kudu employs Multiversion Concurrency Control (MVCC) and the Raft consensus algorithm [4]. Each write operation in Kudu must go through the tablet’s leader.

All replicas of a tablet observe the same order of operations and if a write operation is assigned timestamp n and changes row x, a second write operation at timestamp m > n is guaranteed to see the new value of x.

This strict ordering of lock acquisition and timestamp assignment is enforced to be consistent across all replicas of a tablet through consensus. Therefore, write operations are totally ordered with regard to clock-assigned timestamps, relative to other writes in the same tablet. In other words, writes have strict-serializable semantics, though in an admittedly limited context. See this blog post for a little more context regarding what these semantics mean.

While Isolated and Durable in an ACID sense, write operations are not yet fully Atomic. The failure of a single write in a batch operation does not roll back the operation, but produces per-row errors.

Kudu does not yet support transactions that span multiple tablets. However, consistent snapshot reads are possible (with caveats in the current implementation) as explained below.

Writes to a Kudu client are optionally buffered in memory until they are flushed and sent to the server. During the client’s session flush, the rows for each tablet are batched together, and sent to the tablet server which hosts the leader replica of the tablet. Since there are no inter-tablet transactions, each of these batches represents a single, independent write operation with its own timestamp. However you have the option to impose some constraints on the assigned timestamps and on how writes to different tablets can be observed by clients.

Kudu, like Spanner, was designed to be externally consistent [5], preserving consistency even when operations span multiple tablets and even multiple data centers. In practice this means that, if a write operation changes item x at tablet A, and a following write operation changes item y at tablet B, you might want to enforce that if the change to y is observed, the change to x must also be observed. There are many examples where this can be important. For example, if Kudu is storing clickstreams for further analysis, and two clicks follow each other but are stored in different tablets, subsequent clicks should be assigned subsequent timestamps so that the causal relationship between them is captured.

Kudu’s default external consistency mode is called CLIENT_PROPAGATED. See [1] for an extensive explanation on how it works. In brief, this mode causes writes from a single client to be automatically externally consistent. In this mode, writes are only externally consistent from the perspective of a single client. In the clickstream scenario above, if the two clicks are submitted by different client instances, the application must manually propagate timestamps from one client to the other for the causal relationship to be captured.

CLIENT_PROPAGATED consistency is currently only available on the java client and is exposed through the AsyncKuduClient#getLastPropagatedTimestamp() and AsyncKuduClient#setLastPropagatedTimestamp() methods.

Kudu also implements an experimental implementation of an external consistency model used in Google’s Spanner , called Commit Wait. Commit Wait works by tightly synchronizing the clocks on all machines in the cluster. Then, when a write occurs, timestamps are assigned and the results of the write are not made visible until enough time has passed so that no other machine in the cluster could possibly assign a lower timestamp to a following write.

For the moment, Kudu’s experimental implementation of Commit Wait is only available in the java client, by setting KuduSession#setExternalConsistencyMode() to COMMIT_WAIT. When using this mode, the latency of writes is tightly tied to how accurate the clocks on all the cluster nodes are, and using this mode with loose clock synchronization causes writes to take a long time to complete or even time out. See Known Issues and Limitations.

Scans are read operations performed by clients that may span one or more rows across one or more tablets. When a server receives a scan, it takes a snapshot of the MVCC state and then proceeds in one of two ways depending on the read mode selected by the user by means of the KuduScanner::SetReadMode() method.

Selecting between read modes requires balancing the trade-offs and making a choice that fits your workload. For instance, a reporting application that needs to scan the entire database might need to perform careful accounting operations, so that scan may need to be fault-tolerant, but probably doesn’t require a to-the-microsecond up-to-date view of the database. In that case, you might choose 'READ_AT_SNAPSHOT' and select a timestamp that is a few seconds in the past when the scan starts. On the other hand, a machine learning workload that is not ingesting the whole data set and is already statistical in nature might not require the scan to be repeatable, so you might choose READ_LATEST instead.

We plan to fix the following issues. Monitor the linked JIRAs for progress.

There are several gaps and corner cases that prevent Kudu from being fully strictly-serializable in some situations, at the moment. Below are the details and next, some recommendations.