header

Short Bio

I am a visiting assistant professor at Purdue University. I received my PhD in Computer Science from Pierre and Marie Curie University (Paris 6) under the supervision of Marc Shapiro. My research focus is on systems and algorithms for:


Conference & Workshop

G-DUR: A Middleware for Assembling, Analyzing, and Improving Transactional Protocols.
Masoud Saeida Ardekani, Pierre Sutra, Marc Shapiro.
To appear in 15th Middleware conference (MIDDLEWARE). Bordeaux, France, December 2014.

Abstract: A large family of distributed transactional protocols have a common structure, called Deferred Update Replication (DUR). DUR provides dependability by replicating data, and performance by not re-executing transactions but only applying their updates. Protocols of the DUR family differ only in behaviors of few generic functions. Based on this insight, we offer a generic DUR middleware, called G-DUR, along with a library of finely-optimized plug-in implementations of the required behaviors. This paper presents the middleware, the plugins, and an extensive experimental evaluation in a geo-replicated environment. Our empirical study shows that:(i) G-DUR allows developers to implement various transactional protocols with less than 600 lines of code; (ii) It provides a fair, apples-to-apples comparison between transactional protocols; (iii) By replacing plugs-ins, developers can use G-DUR to understand bottlenecks in their protocols; (iv) This in turn enables the improvement of existing protocols; and (v) Given a protocol, G-DUR allows to evaluate the cost of ensuring various degrees of dependability.

A Self-Configurable Geo-Replicated Cloud Storage System.
Masoud Saeida Ardekani, Douglas B. Terry
In 11th Symposium on Operating Systems Design and Implementation (OSDI). Broomfield, CO, October 2014.

[bib] [pdf] [@UsenixLink]
Abstract: Reconfiguring a cloud storage system can improve its overall service. Tuba is a geo-replicated key-value store that automatically reconfigures its set of replicas while respecting application-defined constraints so that it adapts to changes in clients' locations or request rates. New replicas may be added, existing replicas moved, replicas upgraded from secondary to primary, and the update propagation between replicas adjusted. Tuba extends a commercial cloud-based service, Microsoft Azure Storage, with broad consistency choices (as in Bayou), consistency-based SLAs (as in Pileus), and a novel replication configuration service. Compared with a system that is statically configured, our evaluation shows that Tuba increases the reads that return strongly consistent data by 63%.

Non-Monotonic Snapshot Isolation: scalable and strong consistency for geo-replicated transactional systems.
Masoud Saeida Ardekani, Pierre Sutra, Marc Shapiro.
In 32nd International Symposium on Reliable Distributed Systems (SRDS). Braga, Portugal, October 2013.

[bib] [pdf] [@IeeexploreLink]
Abstract: Modern cloud systems are geo-replicated to improve application latency and availability. Transactional consistency is essential for application developers, however, the corresponding concurrency control and commitment protocols are costly in a geo-replicated setting. To minimize this cost, we identify the following essential scalability properties: (i) only replicas updated by a transaction T make steps to execute T, (ii) a read-only transaction never waits for concurrent transactions and always commits, (iii) a transaction may read object versions committed after it started, and (iv) two transactions synchronize with each other only if their writes conflict. We present Non-Monotonic Snapshot Isolation (NMSI), the first strong consistency criterion to allow implementations with all four properties. We also present a practical implementation of NMSI called Jessy, which we compare experimentally against a number of well-known criteria. Our measurements show that the latency and throughput of NMSI are comparable to the weakest criterion, read-committed, and between two to fourteen times faster than well-known strong consistencies.

On the scalability of Snapshot Isolation.
Masoud Saeida Ardekani, Pierre Sutra, Marc Shapiro, Nuno Preguica.
In 19th International European Conference on Parallel and Distributed Computing (Euro-Par). Aachen, Germany, August 2013.

[bib] [pdf] [@SpringerLink]
Abstract: Many distributed applications require transactions. However, transactional protocols that require strong synchronization are costly in large scale environments. Two properties help with scalability of a transactional system: genuine partial replication (GPR), which leverages the intrinsic parallelism of a workload, and snapshot isolation (SI), which decreases the need for synchronization. We show that under standard assumptions (data store accesses are not known in advance, and transactions may access arbitrary objects in the data store), it is impossible to have both SI and GPR. Our impossibility result is based on a novel decomposition of SI which proves that, like serializability, SI is expressible on plain histories.

The Space Complexity of Transactional Interactive Reads.
Masoud Saeida Ardekani, Marek Zawirski, Pierre Sutra, Marc Shapiro.
In 1st Workshop on Hot Topics in Cloud Data Processing (in conjunction with Eurosys 2012). Bern, Switzerland, April 2012.

[bib] [@acmLink]
Abstract: Many distributed applications require transactions. However, transactional protocols that require strong synchronization are costly in large scale environments. Two properties help with scalability of a transactional system: genuine partial replication (GPR), which leverages the intrinsic parallelism of a workload, and snapshot isolation (SI), which decreases the need for synchronization. We show that under standard assumptions (data store accesses are not known in advance, and transactions may access arbitrary objects in the data store), it is impossible to have both SI and GPR. Our impossibility result is based on a novel decomposition of SI which proves that, like serializability, SI is expressible on plain histories.

The Impossibility of Ensuring Snapshot Isolation in Genuine Replicated STMs.
Masoud Saeida Ardekani, Pierre Sutra, and Marc Shapiro.
In 3rd Workshop on the Theory of Transactional Memory (in conjunction with DISC 2011). Rome, Italy, September 2011.

[bib] [pdf]
Abstract: Many distributed applications require transactions. However, transactional protocols that require strong synchronization are costly in large scale environments. Two properties help with scalability of a transactional system: genuine partial replication (GPR), which leverages the intrinsic parallelism of a workload, and snapshot isolation (SI), which decreases the need for synchronization. We show that under standard assumptions (data store accesses are not known in advance, and transactions may access arbitrary objects in the data store), it is impossible to have both SI and GPR. Our impossibility result is based on a novel decomposition of SI which proves that, like serializability, SI is expressible on plain histories.

TR & Thesis

Ensuring Consistency in Partially Replicated Data Stores.
Masoud Saeida Ardekani.
PhD Thesis, Pierre and Marie Curie University (Paris 6), September 2014.

[bib] [pdf]
Abstract: Cloud-based applications, such as social networking or eCommerce, require to replicate data across several sites to provide responsiveness, availability, and disaster tolerance. Ensuring consistency over a large scale system with slow, and failure prone WANs has become of a paramount importance. This thesis studies this issue. In the first part, we study consistency in a transactional systems, and focus on reconciling scalability with strong transactional guarantees. We identify four scalability properties as being critical for scalability: (i) only replicas updated by a transaction T make steps to execute T; (ii) a read-only transaction never waits for concurrent transactions and always commits; (iii) a transaction may read versions committed after it started; and (iv) two transactions synchronize with each other only if their writes conflict. We show that none of the strong consistency criteria ensure all four. We define a new scalable consistency criterion called Non-Monotonic Snapshot Isolation (NMSI), while is the first that is compatible with all four properties. We also present a practical implementation of NMSI, called Jessy, which we compare experimentally against a number of well-known criteria. Our last contribution in the first part is a framework for performing fair, and apples-to-apples comparison among different transactional protocols. Our insight is that a large family of distributed transactional protocols have a common structure, called Deferred Update Replication (DUR). Protocols of the DUR family differ only in behaviors of few generic functions. We present a generic DUR framework, called G-DUR, along with a library of finely-optimized plug-in implementations of the required behaviors. Our empirical study shows that: (i) G-DUR allows developers to implement various transactional protocols in less than few hundreds lines of code; (ii) It provides a fair, apples-to-apples comparison between transactional protocols; (iii) By replacing plugs-ins, developers can use G-DUR to understand bottlenecks in their protocols; (iv) This in turn enables the improvement of existing protocols; and (v) Given a protocol, G-DUR allows to evaluate the cost of ensuring various degrees of dependability. In the second part of this thesis, we focus on ensuring consistency in non-transactional data stores. We introduce Tuba, a replicated key-value store that dynamically selects replicas in order to maximize the utility delivered to read operations according to a desired consistency defined by the application. In addition, unlike current systems, it automatically reconfigures its set of replicas while respecting application-defined constraints so that it adapts to changes in clients’ locations or request rates. We implemented Tuba on top of Windows Azure Storage (WAS). While providing similar API, Tuba extends WAS with a broad set of consistency choices, consistency-based SLAs, and a geo-replication configuration service. Compared with a system that is statically configured, our evaluation shows that Tuba increases the reads that return strongly consistent data by 63% and improves average utility up to 18%.

Non-Monotonic Snapshot Isolation.
Masoud Saeida Ardekani, Pierre Sutra, Nuno Preguica, Marc Shapiro.
Technical Report RR-7805. INRIA. June 2013.

[bib] [pdf] [@arxivLink]
Abstract: Many distributed applications require transactions. However, transactional pro- tocols that require strong synchronization are costly in large scale environments. Two properties help with scalability of a transactional system: genuine partial replication (GPR), which leverages the intrinsic parallelism of a workload, and snapshot isolation (SI), which decreases the need for synchronization. We show that, under standard assumptions (data store accesses are not known in advance, and transactions may access arbitrary objects in the data store), it is impossible to have both SI and GPR. To circumvent this impossibility, we propose a weaker consistency criterion, called Non-Monotonic Snapshot Isolation (NMSI). NMSI retains the most important properties of SI, i.e., read-only transactions always commit, and two write-conflicting updates do not both commit. We present a GPR protocol that ensures NMSI, and has lower message cost (i.e., it contacts fewer replicas and/or commits faster) than previous approaches.

Making Structured Overlay Networks Data Center Friendly.
Masoud Saeida Ardekani.
Master's Thesis (TRITA-ICT-EX-2010:214), The Royal Institute of Technology (KTH). September 2010.

Most existing overlay networks do not consider deployment over multiple data centers where peers in a same data center are connected with low latency and high bandwidth, while peers in different data centers are connected with high latency and medium bandwidth. De- ploying these systems on data center infrastructures will lead to perfor- mance degradation in lookup latency, and inter-datacenter bandwidth consumption. In this dissertation, we introduce a new class of overlay networks called data center friendly overlays, and characterize their most im- portant properties. We also propose a novel gossip generated inter- datacenter overlay that can be used on top of any SON in order to make it data center friendly. Our solution does not need any modifica- tion to routing, and maintenance protocols of the underlying SON. It uses the Gradient topology to select the highest utility nodes in each DC. These highest utility nodes gossip with nodes in other DCs, and build a gossip generated inter-datacenter overlay. We also propose a novel broadcast solution using the Gradient topology to disseminate gossiped views inside each DC. We also compare our solution performance under different scenar- ios with Kademlia DHT. Our experiments show that by using inter- datacenter overlay on top of a SON, inter-datacenter lookup hops re- duce to one while inter-datacenter traffic is kept minimum. Moreover, using the inter-datacenter overlay on top of a SON, total number of hops decreases, and the whole system performs better under churns, and failures.

Contact

      msaeidaa [a-t} purdue.edu
      Work Phone: (765) 49-69419
      Address:
      Purdue University
      Dept. of Computer Science
      305 N. University Street
      West Lafayette, IN 47906
      USA


Research papers by Masoud Saeida Ardekani SBU KTH SICS UPMC Lip6 MSR INRIAPurdue