CustomFort writes: The HyperDex guys have a new release (1.6), and they're boasting API compatibility with MongoDB, but with a 1-4X speedup vs Mongo (plus strong consistency). Their blog has the details, as well as some code snippets showing off the API matching.
rescrv writes: For the last decade, the
SQL vs. NoSQL argument has been a popular topic of discussion for
developers across the Internet. These discussions typically revolve
vs. BASE or
a discussion of the merits of eventual
consistency. Most NoSQL systems advertise that their systems offer
eventual consistency as a feature, as if it is somehow
desirable to give up on solving the hard problems of distributed
computing. But is giving up on ACID and settling for eventual
consistency the right approach? Now that NoSQL systems have evolved to
the point that there are systems offering strong
consistency guarantees, it's time to revisit the decision to abandon
A new NoSQL system called Warp
offers the strong ACID guarantees of traditional SQL systems with the
scalability and performance of NoSQL systems. Warp employs a new
technique called linear transactions (PDF)
to offer high performance transactions at scale. Preliminary benchmark results are
impressive and show that Warp is comparable to HyperDex, the fastest
key-value store available for comparison.
With it's unique design, high performance, and strong guarantees,
Warp leads us to ask: Who says NoSQL means no ACID?
rescrv writes: Cornell researchers just announced HyperDex Warp, the first key-value store to provide fully-distributed ACID guarantees. Warp enables applications to perform updates on multiple objects in a transactional manner. Is this the first step in a trend where NoSQL bridges the gap to RDBMSes?
denizalti writes: It's difficult to build and deploy distributed systems. Replicating components for fault-tolerance requires mastering a system like ZooKeeper (or Chubby if you're at Google). A new open source system called OpenReplica aims to make this a lot simpler for regular developers.
RemyBR writes: "One month into the Oracle v. Google judgement, judge Alsup said this to Oracle's attorney David Boies: "You're one of the best lawyers in America. I don't know how you could make that argument", in response to Boies' claim that the tiny amount of computer code Google has been found liable for infringing helped it get the Android mobile operating system to market sooner, therefore Oracle should be entitled to a slice of the profits. He then proceeded to reveal his own personal knowledge of the technology in question. Alsup said he has personally written computer code, not in the Java language involved in the lawsuit, but in other languages. And rangeCheck, he said of the nine lines of infringed Java code that Google said it mistakenly put in a version of Android, "is so simple." — "I could do it. You could do it," the judge told Boies. "It was an accident.""
When keys are concatenated with a joining symbol, objects can only be retrieved when one posesses all of the joining keys. Hyperspace hashing allows object retrieval when only a subset of the attributes are available.
Although the coordinator is logically centralized, we've got a version in the works that uses Paxos (a consensus algorithm) to distribute the coordinator as well.
For more information check out http://openreplica.org/
rescrv writes: A recent NoSQL system from Cornell shows that web applications no longer have to trade off strong consistency to achieve high performance. Called HyperDex, the system offers strong consistency, scalability, high performance and fault tolerance. Benchmarks show that HyperDex can achieve speedups of 12-14x over Cassandra, MongoDB and Redis.
judgecorp writes: "Microsoft has filed a complaint with the European Commission complaining that Motorola Mobility is charging too much for use of its patented technology in phones and tablets. The complaint follows a similar one by Apple last week, and will need to be rewolved by Google as it takes charge of Motorola Mobility"
Actually, HyperDex is strongly consistent. A GET will *always* see the most recent PUT. There is no "eventually" in HyperDex. HyperDex's speed advantage stems from a unique, efficient design and a streamlined implementation.
rescrv writes: A new key-value store from Cornell University is set to begin a new era of NoSQL storage. The system, called HyperDex, enables efficient searches over the stored values, while retaining the traditional get/put interface of a key-value store. HyperDex provides significant performance advantages over Cassandra and MongoDB for both traditional key-value operations, and for search.
rescrv writes: Key-value stores (like Cassandra, Redis and DynamoDB) have been replacing traditional databases in many demanding web applications (e.g. Twitter, Google, Facebook, LinkedIn and others). But for the most part, the differences between existing NoSQL systems come down to the choice of well-studied implementation techniques; in particular, they all provide a similar API that achieves high performance and scalability by limiting applications to simple operations like GET and PUT.
HyperDex, a new key-value store developed at Cornell, stands out in the NoSQL spectrum with its unique design. HyperDex employs a unique multi-dimensional hash function to enable efficient search operations — that is, objects may be retrieved without using the key under which they are stored. Other systems employ indexing techniques to enable search, or enumerate all objects in the system. In contrast, HyperDex's design enables applications to retrieve search results directly from servers in the system. The results are impressive. Preliminary benchmark results on the project website show that HyperDex provides significant performance improvements over Cassandra and MongoDB. With its unique design, and impressive performance, it seems fittng to ask: Is HyperDex the start of NoSQL 2.0?
An anonymous reader writes: Many people (myself included!) have found that moving from an RDBMS to a NoSQL solution in search of better performance is often a long and unrewarding exercise. Secondary index support is still hit and miss with most NoSQL systems. So unless you are willing to spend the time and effort to re-normalize your schemas the NoSQL way, you are unlikely to see any performance benefit from using NoSQL. Even if you do re-normalize, you still have to change your application to handle weak eventual consistency guarantees that is standard with NoSQL solutions.
A new system from Cornell is trying to bridge the performance/functionality gap between NoSQL and RDBMS. It is called HyperDex and it uses "hyperspace hashing" to allow efficient searches on secondary attributes without requiring explicit indices. It also provides strong consistency for most of its operations. The developers have posted some performance numbers that completely blow away Cassandra and MongoDB. Have we finally found the sweet spot in the NoSQL/RBDMS design space?
el33thack3r writes: NoSQL systems now power many websites (including Slashdot) because they offer high performance and scalability. These properties usually come at the expense of functionality — NoSQL stores typically support a very limited GET/PUT interface.
A new announcement from Cornell researchers show that NoSQL stores are beginning to offer functionality found in traditional SQL systems, without compromising the high performance and scalability that have been the hallmarks of the NoSQL movement. The new high performance key-value store, called HyperDex, supports efficient SEARCH operations. HyperDex uses a novel object placement strategy, called hyperspace hashing, such that the key-value store can retrieve objects not just by the primary key, but any of the object's secondary attributes as well. Hyperspace hashing ensures that both search and key-based operations contact a small subset of all servers in a cloud deployment. Performance results show that the system is a factor of 2-13 times faster than Cassandra and MongoDB.
rescrv writes: Researchers at Cornell have built a new high performance key-value store called HyperDex which supports efficient search operations. HyperDex uses a careful object placement strategy, called hyperspace hashing, such that the key-value store can retrieve objects not just by the primary key, but any of the object's secondary attributes as well. Hyperspace hashing ensures that both search and key-based operations contact a small subset of all servers in the system. Performance results show that the system is a factor of 2-13 times faster than Cassandra and MongoDB.