Subscribe Free M2M Magazine
IoT WORLD FORUM 2017

ParStream launches IoT Analytics Platform

| November 5, 2014

PARSTREAM IOT ANALYTICS

ParStream has announced the launch of the ParStream Analytics Platform, designed to analyze data produced by the Internet of Things (IoT). With the ParStream DB database at its core, the Analytics Platform reportedly can analyze billions of rows of data and still deliver sub-second response times.

IoT WORLD FORUM 2014

The 451 Take

ParStream already had an impressive columnar database that could not only handle real-time analytics on vast amounts of events, but also had a small footprint to boot thanks to its Compressed Index. With the launch of the ParStream Analytics Platform, the company has now really set up shop to sell into the IoT space. While it still relies on partnerships to provide all of the pieces of the puzzle, ParStream is further down the road with an integrated stack than some, and it will likely appeal to those seeking both performance and fast time to market.

Context

Cologne-based ParStream was founded in 2009 to develop databases able to handle large volumes of real-time queries. The company was started because the founders – CEO Mike Hummel and CTO Jörg Bienert – were struggling to find suitable technology to support the development of an online travel portal in their previous roles.

ParStream has raised $18.3m in total funding thus far from Khosla Ventures, Baker Capital, CrunchFund, Data Collective, Tola Capital and some private investment. The company has about 45 staff – nearly all in engineering – up from 39 in October 2013, and roughly a dozen customers in production, compared with nine in October 2013.

ParStream believes its database technology solves a problem that can’t be handled by other databases. The company argues that today, more use cases require real-time data import, real-time data querying and flexible analytics. Traditional relational databases aren’t fast enough or scalable enough (at least not affordably); Hadoop is limited by its batch response times; and key value stores have limited analytic capabilities because they are harder to integrate with traditional BI and reporting tools.

As such, ParStream built a distributed, massively parallel processing columnar database based on a shared-nothing architecture. Perhaps key to its potential is the company’s patented High Performance Compressed Index, which keeps its footprint down. Its lockless architecture enables fast querying and concurrent high-bandwidth data import – traditional relational databases that rely on locking in order to keep records consistent are not able to ingest data and query it concurrently at anywhere near the same speed as ParStream, according to the firm.

Besides, most businesses like to keep analytics on their transactional databases to a minimum to safeguard the performance of their enterprise applications. Perhaps the real key to ParStream’s differentiation is the ability to analyze data while it is still in its compressed state, rather than having to decompress it first (which would add latency).

Technology

While the company already had the ParStream DB database – which it said in May would make an ideal platform for those starting to try and tap into the IoT – it lacked a suite of tools to make it useful to anyone other than developers. All of that has changed with the release of the ParStream Analytics Platform.

The platform is in limited release now, with general availability expected by the end of this year. It is priced based on the amount of data being analyzed, but the exact pricing has not been set yet.

The platform queries at the source of data for real-time analysis as data is being loaded. It reportedly provides unified analytics of real-time data in every query, and generates more accurate insights for decision-makers with the continuous import of new data. Importantly, ParStream says the platform can be used to query historical as well as real-time data.

The company has pulled out all the stops in trying to deliver an IoT platform, rather than just a couple of bolted-on apps that sit on its database. In addition to ParStream DB, the ParStream Analytics Platform includes:

  • Geo-distributed analytics that enable the analytics to be run across remote data sources and data close to the data source.
  • Time-series analytics that enable enterprises to analyze streaming data together with massive amounts of historical data.
  • Alerts and actions to monitor data streams, create and qualify alerts, and appropriately execute actions automatically or send notifications.
  • Advanced analytics to derive models and hypotheses from large amounts of data by applying statistical functions and analytical models.
  • ParStream DB integrates with KNIME, statistical libraries like R, and other machine-learning engines.

However, the company acknowledges that even now it doesn’t have all of the pieces of the puzzle. To that end, it has integrated with Datawatch for its front-end visualization tool. It says it can integrate with others such as Tableau, Qlik, MicroStrategy and Yellowfin but these aren’t certified integrations for the Analytics Platform. It also offers integration with Informatica’s collection tool, Vibe, for the high-velocity ingestion needs of the IoT. Taken together, ParStream believes businesses can expect to roll out IoT applications twice as fast as integrating components in-house from different vendors.

Customers

ParStream doesn’t yet have any reference customers for the Analytics Platform, but it cites organizations including MPREIS, Cake and INRA as ParStream DB users. MPREIS, an Austria-based supermarket chain, was using QlikView to analyze its point-of-sale data, but it had hit a database bottleneck at 400 million records. By using ParStream with QlikView, the company can now analyze up to 30 billion records effectively. In a use case that it says paves the way for other IoT scenarios, the company reports that a telecom provider (which it can’t yet reference by name) deployed the technology to continuously import more than one million rows per second per node.

Competition

As a massively parallel columnar database, ParStream can be compared with the likes of SAP’s Sybase IQ and HP’s Vertica. We also put Infobright in the columnar database camp, particularly since it also markets its technology toward the IoT.

We’d also expect ParStream to face competition from SAP’s in-memory HANA database and Amazon Web Services’ Redshift (which is, in turn, based on Actian’s Matrix database), depending on the situation. The company is not unique in its ability to analyze data while it is still in its compressed state. Indeed, its approach reminds us of RainStor and its combination of columnar storage and compression, which provides the ability to perform SQL queries against compressed data.

In the wider data-warehousing sector, ParStream faces competition from the likes of Oracle, Microsoft, Teradata, IBM and Kognitio. Depending on the use case, ThoughtSpot and jKool are among a group of emerging players promising analytics with much greater ease of use, thanks to natural language-like text search of large datasets.

Depending on one’s definition of the IoT, ParStream may find itself coming up against machine data analytic firms such as Splunk, Loggly, Logentries and X15 Software. InfiniDB, which was also in this space, closed shop in October, citing stiff competition as the reason for its demise.

SWOT Analysis

Strengths Weaknesses
ParStream has a highly performing underlying database suited to the rapid analysis of streaming events, and as such fitting for the IoT, which requires both speed and scale. The additional elements on top are starting to feel much more like a platform rather than a database alone. The platform isn’t complete just yet: visualization and data preparation are still handled by integrations with partners (primarily DataWatch and Informatica). As a startup, ParStream has relatively few reference customers.
Opportunities Threats
We’ve said before that we think more businesses will ask how the Internet of Things, or at least the Internet of Their Things, could play a role in their analytic and decision-making processes in the next two years. But we do now need to start to see real-world examples rather than mere hype. ParStream is still a young company. This space is incredibly competitive, with the huge incumbents battling it out with plenty of startups both on-premises and in the cloud, and one rival (InfiniDB) declared bankruptcy this year.

inShare

 

 

This report falls under the following categories. Click on a link below to find similar documents.

Company: ParStream 
Other Companies: Actian CorpAmazon Web ServicesBaker CapitalCakeCrunchFundData CollectiveDatawatchHPIBM,InfiniDBInfobrightInformaticaINRAjKoolKhosla VenturesKNIMEKognitioLogentriesLogglyMicrosoftMicroStrategy,MPREISOracleQlikRainStorSAPSplunkSybaseTableauTeradataThoughtSpotTola CapitalVertica SystemsX15 SoftwareYellowfin

Analyst(s): Jason Stamper

 

Tags: , , , , , ,

Category: NEWS

Comments are closed.

Subscribe Free M2M Magazine

Subscribe FREE "M2M Magazine"

* indicates required