next up previous
Next: Evaluation Up: Beehive: O(1) Lookup Performance Previous: Mutable Objects


Implementation

Beehive is a general replication mechanism that can be applied to any prefix-based distributed hash table. We have layered our implementation on top of Pastry, a freely available DHT with logN lookup performance. Our implementation is structured as a transparent layer on top of FreePastry 1.3, supports a traditional insert/modify/delete/query DHT interface for applications, and required no modifications to underlying Pastry. However, converting the preceding discussion into a concrete implementation of the Beehive framework, building a DNS application on top, and combining the framework with Pastry required some practical considerations and identified some optimization opportunities.

Beehive needs to maintain some additional, modest amount of state in order to track the replication level, freshness, and popularity of objects. Each Beehive node stores all replicated objects in an object repository. Beehive associates the following meta-information with each object in the system, and each Beehive node maintains the following fields within each object in its repository:

In addition to the state associated with each object, Beehive nodes also maintain a running estimate of the Zipf parameter. Overall, the storage cost consists of several bytes per object, and the processing cost of keeping the meta-data up to date is small.

Pastry's query routing deviates from the model described earlier in the paper because it is not entirely prefix-based and uniform. Since Pastry maps each object to the numerically closest node in the identifier space, it is possible for an object to not share any prefixes with its home node. For example, in a network with two nodes 298 and 315, Pastry will store an object with identifier 304 on node 298. Since a query for object 304 propagated by prefix matching alone cannot reach the home node, Pastry completes the query with the aid of an auxiliary data structure called leaf set. The leaf set is used in the last few hops to directly locate the numerically closest node to the queried object. Pastry initially routes a query using entries in the routing table, and may route the last couple of hops using the leaf set entries. This required us to modify Beehive's replication protocol to replicate objects at the leaf set nodes as follows. Since the leaf set is most likely to be used for the last hop, we replicate objects in the leaf set nodes only at the highest replication levels. Let k = logbN be the highest replication level for Beehive, that is, the default replication level for an object replicated only at its home node. As part of the replicate phase, a node A sends a replication message to all nodes B in its routing table as well as its leaf set with a list of identifiers of objects replicated at level k-1 whose deciding node is B. B is the deciding node of an object homed at node A, if A would forward a query to that object to node B next. Upon receiving a maintenance message at level k-1, node B would push an object to node A only if node A and the object have at least k-1 matching prefixes. Once an object is replicated on a leaf set node at level k-1, further replication to lower levels follow the replication protocol described in Section 2. This slight modification to Beehive enables it to work on top of Pastry. Other routing metrics for DHT substrates, such as the XOR metric [18], have been proposed that do not exhibit this non-uniformity, and where the Beehive implementation would be simpler.

Pastry's implementation provides two opportunities for optimization, which improve Beehive's impact and reduce its overhead. First, Pastry nodes preferentially populate their routing tables with nodes that are in physical proximity [4]. For instance, a node with identifier 100 has the opportunity to pick either of two nodes 200 and 201 when routing based on the first digit. Pastry selects the node with the lowest network latency, as measured by the packet round-trip time. As the prefixes get longer, node density drops and each node has progressively less freedom to find and choose between nearby nodes. This means that a significant fraction of the lookup latency experienced by a Pastry lookup is incurred on the last hop. Hence, selecting even a large number of constant hops, C, as Beehive's performance target, will have a significant effect on the real performance of the system. While we pick C=1 in our implementation, note that C is a continuous variable and may be set to a fractional value, to get average lookup performance that is a fraction of a hop. C=0 yields a solution that will replicate all objects at all hops, which is suitable only if the total hash table size is small.

The second optimization opportunity stems from the periodic maintenance messages used by Beehive and Pastry. Beehive requires periodic communication between nodes and the member of their routing table and leaf-set for replica dissemination and data aggregation. Pastry nodes periodically send heart-beat messages to nodes in their routing table and leaf set to detect node failures. They also perform periodic network latency measurements to nodes in their routing table in order to obtain closer routing table entries. We can improve Beehive's efficiency by combining the periodic heart-beat messages sent by Pastry with the periodic aggregation messages sent by Beehive. By piggy-backing the ith row routing table entries on to the Beehive aggregation message at replication level i, a single message can simultaneously serve as a heart beat message, Pastry maintenance message, and a Beehive aggregation message.

We have built a prototype DNS name server on top of Beehive in order to evaluate the caching strategy proposed in this paper. Beehive-DNS uses the Beehive framework to proactively disseminate DNS resource records containing name to IP address bindings. The Beehive-DNS server currently supports UDP-based queries and is compatible with widely-deployed resolver libraries. Queries that are not satisfied within the Beehive system are looked up in the legacy DNS by the home node and are inserted into the Beehive framework. The Beehive system stores and disseminates resource records to the appropriate replication levels by monitoring the DNS query stream. Clients are free to route their queries through any node that is part of the Beehive-DNS. Since the DNS system relies entirely on aggressive caching in order to scale, it provides very loose coherency semantics, and limits the rate at which updates can be performed. Recall that the Beehive system enables resource records to be modified at any time, and disseminates the new resource records to all caching name servers as part of the update operation. However, for this process to be initiated, name owners would have to directly notify the home node of changes to the name to IP address binding. We expect that, for some time to come, Beehive will be an adjunct system layered on top of legacy DNS, and therefore name owners who are not part of Beehive will not know to contact the system. For this reason, our current implementation delineates between names that exist solely in Beehive versus resource records originally inserted from legacy DNS. In the current implementation, the home node checks for the validity of each legacy DNS entry by issuing a DNS query for the domain when the time-to-live field of that entry is expired. If the DNS mapping has changed, the home node detects the update and propagates it as usual. Note that this strategy preserves DNS semantics and is quite efficient because only the home nodes check the validity of each entry, while replicas retain all mappings unless invalidated.

Overall, the Beehive implementation adds only a modest amount of overhead and complexity to peer-to-peer distributed hash tables. Our prototype implementation of Beehive-DNS is only 3500 lines of code, compared to the 17500 lines of code for Pastry.


next up previous
Next: Evaluation Up: Beehive: O(1) Lookup Performance Previous: Mutable Objects
beehive-l@cs.cornell.edu 2004-02-11