Cristiana Amza

TreadMarks: shared memory computing on networks of workstations

Shared memory facilitates the transition from sequential to parallel processing. Since most data structures can be retained, simply adding synchronization achieves correct, efficient programs for many applications. We discuss our experience with parallel computing on networks of workstations using the TreadMarks distributed shared memory system. DSM allows processes to assume a globally shared virtual memory even though they execute on nodes that do not physically share memory. We illustrate a DSM system consisting of N networked workstations, each with its own memory. The DSM software provides the abstraction of a globally shared memory, in which each processor can access any data item without the programmer having to worry about where the data is or how to obtain its value.

Exploiting distributed version concurrency in a transactional memory cluster

We investigate a transactional memory runtime system providing scaling and strong consistency, i.e., 1-copy serializability on commodity clusters for both distributed scientific applications and database applications. We introduce a novel page-level distributed concurrency control algorithm, called Distributed Multiversioning (DMV). DMV automatically detects and resolves conflicts caused by data races for distributed transactions accessing shared in-memory data structures. DMVs key novelty is in exploiting the distributed data versions that naturally come about in a replicated cluster in order to avoid read-write conflicts, hence provide scaling. DMV runs conflicting read-only and update transactions in parallel on different replicas instead of using different physical data copies within a single node as in classic multiversioning. In its most general form, DMV can be used to implement a software transactional memory system on a cluster for scaling C++ applications. DMV supports highly multithreaded database applications as well by centralizing updates on a master replica and creating the required page versions for read-only transactions lazily, on a set of slave replicas. We also show that a version-aware scheduling technique can distribute the read-only transactions across the slaves in such a way to minimize version conflicts.In our evaluation, we use DMV as a lightweight approach to scaling a hash table microbenchmark workload and the industry-standard e-commerce workload of the TPC-W benchmark on a commodity cluster. Our measurements show scaling for both benchmarks. In particular, we show near-linear scaling up to 8 transactional nodes for the most common e-commerce workload, the TPC-W shopping mix. We further show that our scaling for the TPC-W e-commerce benchmark compares favorably with that of an existing coarse-grained asynchronous replication technique.

Autonomic Provisioning of Backend Databases in Dynamic Content Web Servers

In autonomic provisioning, a resource manager allocates resources to an application, on-demand, e.g., during load spikes. Modelling-based approaches have proved very successful for provisioning the web and application server tiers in dynamic content servers. On the other hand, accurately modelling the behavior of the back-end database server tier is a daunting task. Hence, automated provisioning of database replicas has received comparatively less attention. This paper introduces a novel pro-active scheme based on the classic K-nearest-neighbors (KNN) machine learning approach for adding database replicas to application allocations in dynamic content web server clusters. Our KNN algorithm uses lightweight monitoring of essential system and application metrics in order to decide how many databases it should allocate to a given workload. Our pro-active algorithm also incorporates awareness of system stabilization periods after adaptation in order to improve prediction accuracy and avoid system oscillations. We compare this pro-active self-configuring scheme for scaling the database tier with a reactive scheme. Our experiments using the industry-standard TPC-W e-commerce benchmark demonstrate that the pro-active scheme is effective in reducing both the frequency and peak level of SLA violations compared to the reactive scheme. Furthermore, by augmenting the pro-active approach with awareness and tracking of system stabilization periods induced by adaptation in our replicated system, we effectively avoid oscillations in resource allocation.

Locality aware dynamic load management for massively multiplayer games

Most massively multiplayer game servers employ static partitioning of their game world into distinct mini-worlds that are hosted on separate servers. This limits cross-server interactions between players, and exposes the division of the world to players. We have designed and implemented an architecture in which the partitioning of game regions across servers is transparent to players and interactions are not limited to objects in a single region or server. This allows a finer grain partitioning, which combined with a dynamic load management algorithm enables us to better handle transient crowding by adaptively dispersing or aggregating regions from servers in response to quality of service violations.Our load balancing algorithm is aware of the spatial locality in the virtual game world. Based on localized information, the algorithm balances the load and reduces the cross server communication, while avoiding frequent reassignment of regions. Our results show that locality aware load balancing reduces the average user response time by up to a factor of 6 compared to a global algorithm that does not consider spatial locality and by up to a factor of 8 compared to static partitioning.

Distributed versioning: consistent replication for scaling back-end databases of dynamic content web sites

Dynamic content Web sites consist of a front-end Web server, an application server and a back-end database. In this paper we introduce distributed versioning, a new method for scaling the back-end database through replication. Distributed versioning provides both the consistency guarantees of eager replication and the scaling properties of lazy replication. It does so by combining a novel concurrency control method based on explicit versions with conflict-aware query scheduling that reduces the number of lock conflicts. We evaluate distributed versioning using three dynamic content applications: the TPC-W e-commerce benchmark with its three workload mixes, an auction site benchmark, and a bulletin board benchmark. We demonstrate that distributed versioning scales better than previous methods that provide consistency. Furthermore, we demonstrate that the benefits of relaxing consistency are limited, except for the conflict-heavy TPC-W ordering mix.

Adaptive energy conserving algorithms for neighbor discovery in opportunistic Bluetooth networks

In this paper, we introduce and evaluate novel adaptive schemes for neighbor discovery in Bluetooth-enabled ad-hoc networks. In an ad-hoc peer-to-peer setting, neighbor search is a continuous, hence battery draining process. In order to save energy when the device is unlikely to encounter a neighbor, we adaptively choose parameter settings depending on a mobility context to decrease the expected power consumption of Bluetooth-enabled devices. For this purpose, we first determine the mean discovery time and power consumption values for In different Bluetooth parameter settings through a comprehensive exploration of the parameter space by means of simulation validated by experiments on real devices. The fastest average discovery time obtained is 0.2 s, while at an average discovery time of I s the power consumption is just 1.5 times that of the idle mode on our devices. We then introduce two adaptive algorithms for dynamically adjusting the Bluetooth parameters based on past perceived activity in the ad-hoc network. Both adaptive schemes for selecting the discovery mode are based only on locally-available information. We evaluate these algorithms in a node mobility simulation. Our adaptive algorithms reduce energy consumption by 50% and have up to 8% better performance over a static power-con serving scheme

A comparative evaluation of transparent scaling techniques for dynamic content servers

We study several transparent techniques for scaling dynamic content Web sites, and we evaluate their relative impact when used in combination. Full transparency implies strong data consistency as perceived by the user, no modifications to existing dynamic content site tiers and no additional programming effort from the user or site administrator upon deployment. We study strategies for scheduling and load balancing queries on a cluster of replicated database back-ends. We also investigate transparent query caching as a means of enhancing database replication. Our work shows that, on an experimental platform with up to 8 database replicas, the various techniques work in synergy to improve overall scaling for the e-commerce TPC-W benchmark. We rank the techniques necessary for high performance in order of impact as follows. Key among the strategies are scheduling strategies, such as conflict-aware scheduling, that minimize consistency maintenance overheads. The choice of load balancing strategy is less important. Transparent query result caching increases performance significantly at any given cluster size for a mostly-read workload. Its benefits are limited for write-intensive workloads, where content-aware scheduling is the only scaling option.

Adaptive protocols for software distributed shared memory

We demonstrate the benefits of software shared memory protocols that adapt at run time to the memory access patterns observed in the applications. This adaptation is automatic-no user annotations are required-and does not rely on compiler support or special hardware. We investigate adaptation between singleand multiple-writer protocols, dynamic aggregation of pages into a larger transfer unit, and adaptation between invalidate and update. Our results indicate that adaptation between single- and multiple-writer and dynamic page aggregation are clearly beneficial. The results for the adaptation between invalidate and update are less compelling, showing at best gains similar to the dynamic aggregation adaptation and at worst serious performance deterioration.

Software DSM protocols that adapt between single writer and multiple writer

We present two software DSM protocols that dynamically adapt between a single writer (SW) and a multiple writer (MW) protocol based on the applications sharing patterns. The first protocol (WFS) adapts based on write-write false sharing; the second (WFS+WG) based on a combination of write-write false sharing and write granularity. The adaptation is automatic. No user or compiler information is needed. The choice between SW and MW is made on a per-page basis. We measured the performance of our adaptive protocols on an 8-node SPARC cluster connected by a 155 Mbps ATM network. We used eight applications, covering a broad spectrum in terms of write-write false sharing and write granularity. We compare our adaptive protocols against the MW-only and the SW-only approach. Adaptation to write-write false sharing proves to be the critical performance factor, while adaptation to write granularity plays only a secondary role in our environment and for the applications considered. Each of the two adaptive protocols matches or exceeds the performance of the best of MW and SW in seven out of the eight applications.

Tradeoffs between false sharing and aggregation in software distributed shared memory

Software Distributed Shared Memory (DSM) systems based on virtual memory techniques traditionally use the hardware page as the consistency unit. The large size of the hardware page is considered to be a performance bottleneck because of the implied false sharing overheads. Instead, we show that in the presence of a relaxed consistency model and a multiple writer protocol, a large consistency unit is generally not detrimental to performance. We study the tradeoffs between false sharing and aggregation effects when using large consistency units. In this context, this paper makes three separate contributions:1. We document the cost of false sharing in terms of extra messages and extra data being communicated. We find that, for the applications considered, when the virtual memory page is used as the consistency unit, the number of extra messages is small, while the amount of extra data can be substantial.2. We evaluate the performance when the consistency unit is increased to a multiple of the virtual memory page size. For most applications and data sets, the performance improves, except when the false sharing effects include extra messages or a large amount of extra data.3. We present a new algorithm for dynamically aggregating pages. In our algorithm, the aggregated pages do not necessarily need to be contiguous. In all cases, the performance of our dynamic aggregation algorithm is similar to that achieved with the best static page size.These results were obtained by measuring the performance of eight applications on the TreadMarks distributed shared memory system. The hardware platform used is a network of 166Mhz Pentiums connected by a switched 100Mbps Ethernet network.