Name: 002 - Eventual Consistency & Scalable JVM-based Services at Simple
Start: 2012-02-28T19:00:00-08:00
End: 2012-02-28T22:00:00-08:00
Location: BashoWest

Hi All,

We are set to go for the second installment of BashoChats. As you'll see, this months's lineup is nothing short of spectacular. Per usual, we'll be holding this at BashoWest, located at 929 Market St. The date and time for this month's meetup are February 28 at 7PM, respectively. Be sure to arrive on time as we have a lot to cover...

----------

7:15

How Eventual is Eventual Consistency? Probabilistically Bounded Staleness for Practical Partial Quorums

Peter Bailis (http://www.eecs.berkeley.edu/~pbailis/index.html), UC Berkeley

Many distributed data stores such as Riak, Cassandra, and Voldemort support partial quorum replication. Under partial quorum replication, a data store waits for only a subset of replicas to respond to a request before returning an answer, without ensuring that read and write replica sets intersect (R+W <= N). While these eventually consistent configurations can dramatically improve operation latency compared to strongly consistent configurations, they provide no guarantees regarding the recency of data returned.

In recent work at UC Berkeley (http://www.eecs.berkeley.edu/~pbailis/projects/pbs/), we examined when and why partial quorums return consistent data. In this talk, we introduce Probabilistically Bounded Staleness (PBS), which provides expected bounds on staleness with respect to both versions and wall clock time. Using production data from LinkedIn and Yammer, we quantitatively demonstrate why, in practice, eventually consistent partial quorums often serve consistent data. We will also discuss how to apply PBS to optimize the latency-consistency trade-off inherent in data stores like Riak.

8:00

Instant Service-Oriented Architecture

Ted Nyman (https://twitter.com/#!/tnm), Lead Engineer, Simple (https://simple.com/)

New tools and approaches allow even small engineering teams and early-stage companies to build maintainable, instrumented, fast, ops-friendly, scalable distributed services. I'll talk about the RESTful service-oriented JVM-backed architecture we've build at Simple. I'll discuss new libraries and frameworks like Dropwizard (http://dropwizard.codahale.com/index.html) which allow the development of sophisticated services with surprising little development time, and how SOA can work at any scale.

----------

Seats are limited, so RSVP early and often. See you on February 28th. Let me know if you have any questions.

Mark
mark@basho.com

Mark Phillips

BashoChats

Technology

Database Development

Big Data

NoSQL

Distributed Systems

Riak

Computing

002 - Eventual Consistency & Scalable JVM-based Services at Simple

BashoWest

Share

BashoChats

002 - Eventual Consistency & Scalable JVM-based Services at Simple

BashoChats

Details

Members are also interested in