Wednesday Mar 16, 2016

Maximum Availability Architecture for Big Data Appliance

Oracle Maximum Availability Architecture (MAA) is Oracle's best practices blueprint based on proven Oracle high availability technologies, along with expert recommendations and customer experiences. MAA best practices have been highly integrated into the design and operational capability of Oracle Big Data Appliance, and together they provide the most comprehensive highly available solution for Big Data.

Oracle MAA papers are published at the MAA home page of the Oracle Technology Network (OTN) website. Oracle Big Data Appliance (BDA) Maximum Availability Architecture is a best-practices blueprint for achieving an optimal high-availability deployment using Oracle high-availability technologies and recommendations.

The Oracle BDA MAA exercise for this paper was executed on Oracle Big Data Appliance and Oracle Exadata Database Machine to validate high availability and to measure downtime in various outage scenarios. The current release of this technical paper covers the first phase of the overall Oracle BDA MAA project. The project comprises the following two phases:

Phase 1: High Availability and Outage scenarios at a single site

Phase 2: Disaster Recovery Scenarios across multiple sites

The white paper covering Phase 1 is now published here

Friday Oct 23, 2015

Performance Study: Big Data Appliance compared with DIY Hadoop

Over the past couple of months a team of Intel engineers have been working with our engineers on Oracle Big Data Appliance and performance, especially in ensuring a BDA outperforms DIY Hadoop out of the box. The good news is that your BDA, as you know it today is already 1.2x faster. We are now working to include a lot of the findings in BDA 4.3 and subsequent versions, so we are steadily expanding that 1.2x into a 2x out of box performance advantage. And that is all above and beyond the faster time to value a BDA delivers, as well as on top of the low cost you can get it for. 

Read the full paper here.

But, we thought we should add some color to all of this, and if you are at Openworld this year, come listen to Eric explain all of this in detail on Monday October 26th at 2:45 in Moscone West room 3000.

If you can't make it, here is a short little dialog we had over the results and both Eric and Lucy's take on the work they did and what they are up to next.

Q: What was the most surprising finding in tuning the system?

A: We were surprised how well the BDA performed right after its installation. Having worked for over 5 years on Hadoop, we understand it is a long iterative process to extract the best possible performance out of your hardware. BDA was a well-tuned machine and we were a little concerned we might not have much value to add... 

Q: Anything that you thought was exciting but turned out to be not such a big thing?

A: We were hoping for 5x gains from our work, but only got 2x... But, in all seriousness, we were hoping for better results from some of our memory and Java garbage collection tuning. Unfortunately they only resulted in marginal single digits gains. 

Q: What is next?

A: There is a long list of exciting new products coming from Intel in the coming year; such as hardware accelerated compression, 3d-Xpoint, almost zero latency PCIE SSDs and not to forget new processors. We are excited at the idea of tightly integrating them all with Big Data technologies! What is a better test bed that the BDA? A full software/hardware solution!

Looks like we have a lot of fun things to go work on and with, as well as of course looking into performance improvements for BDA in light of Apache Spark.

See you all at Openworld, or once again, read the paper here

Wednesday Mar 18, 2015

Production workloads blend Cloud and On-Premise Capabilities

Prediction #7 - blending production workloadsacross cloud and on-premise in Oracle's Enterprise Big Data Predictions 2015 is a tough nut to crack. Yet, we at Oracle think this is really the direction we all will go. Sure we can debate the timing, and whether or not this happens in 2015, but it is something that will come to all of us who are looking towards that big data future. So let’s discuss what we think is really going to happen over the coming years in the big data and cloud world.

Reality #1 – Data will live both in the cloud and on-premise

We see this today. Organizations run Human Capital Management systems in the cloud, integrate these with data from outside cloud based systems (think for example LinkedIn, staffing agencies etc.) while their general data warehouses and new big data systems are all deployed as on-premise systems. We also see the example in the prediction where various auto dealer systems uplink into the cloud to enable the manufacturer to consolidate all of their disparate systems. This data may be translated into data warehouse entries and possibly live in two worlds – both in the cloud and on-premise for deep diving analytics or in aggregated form.

Reality #2 – Hybrid deployments are difficult to query and optimize

We also see this today and it is one of the major issues of living in the hybrid world of cloud and on-premise. A lot of the issues are driven by low level technical limitations, specifically in network bandwidth and upload / download capacity into and out of the cloud environment. The other challenges are really (big) data management challenges in that they go into the art of running queries across two ecosystems with very different characteristics. We see a trend to use engineered systems on-premise, which delivers optimized performance for the applications, but in the cloud we often see virtualization pushing the trade-off towards ease of deployment and ease of management. These completely different ecosystems make optimization of queries across them very difficult.

Solution – Equality brings optimizations to mixed environments

As larger systems like big data and data warehouse systems move to the cloud, better performance becomes a key success criterion. Oracle is uniquely positioned to drive both standardization and performance optimizations into the cloud by deploying on engineered systems like Oracle Exadata and Oracle Big Data Appliance. Deploying engineered systems enables customers to run large systems in the cloud delivering performance as they see today in on-premise deployments. This then means that we do not live in a world divided in slow and fast, but in a world of fast and fast.
This equivalence also means that we have the same functionality in both worlds, and here we can sprinkle in some – future – Oracle magic, where we start optimizing queries to take into account where the data lives, how fast we can move it around (the dreaded networking bandwidth issue) and where we need to execute code. Now, how are we going to do this? That is a piece magic, and you will just need to wait a bit… suffice it to say we are hard at work at solving this challenging topic.

Thursday Oct 09, 2014

One of the ways Oracle is using Big Data

Today, Oracle is using big data technology and concepts to significantly improve the effectiveness of its support operations, starting with its hardware support group. While the company is just beginning this journey, the initiative is already delivering valuable benefits.

In 2013, Oracle’s hardware support group began to look at how it could use automation to improve support quality and accelerate service request (SR) resolution. Its goal is to use predictive analytics to automate SR resolution within 80% to 95% accuracy.

Oracle’s support group gathers a tremendous amount of data. Each month, for example, it logs 35,000 new SRs and receives nearly 6 TB of telemetry data via automated service requests (ASRs)—which represent approximately 18% of all SRs. Like many organizations, Oracle had a siloed view of this data, which hindered analysis. For example, it could look at SRs but could not analyze the associated text, and it could review SRs and ASRs separately, but not together.

Oracle was conducting manual root-cause analysis to identify which types of SRs were the best candidates for automation. This was a time-consuming, difficult, and costly process, and the company looked to introduce big data and predictive analytics to automate insight.

The team knew that it had to walk before it could run. It started by taking information from approximately 10 silos, such as feeds from SRs and ASRs, parts of databases, and customer experience systems, and migrating the information to an Oracle Endeca Information Discovery environment. Using the powerful Oracle Endeca solution, Oracle could look at SRs, ASRs, and associated notes in a single environment, which immediately yielded several additional opportunities for automation. On the first day of going live with the solution, Oracle identified 4% more automation opportunities.

Next, Oracle focused its efforts on gaining insight in near real time, leveraging the parallel processing of Hadoop to automatically feed Oracle Endeca Information Discovery—dramatically improving data velocity. Oracle’s first initiative with this new environment looked at Oracle Solaris SRs. In the first few weeks of that project, Oracle identified automation opportunities that will increase automated SR resolution from less than 1% to approximately 5%—simply by aggregating all of the data in near real-time. 

Once Oracle proved via these early proofs of concept that it could process data more efficiently and effectively to feed analytical projects, it began to deploy Oracle Big Data Appliance and Oracle Exalytics In-Memory Machine.

Read the entire profile here.

Monday Sep 29, 2014

The End of the Data Scientist Bubble...

Looking around northern California and inside many technology kitchens makes me believe that we are about to see the Data Scientist bubble burst. And then I read the Fortune Magazine article on Peter Thiel - and the excerpt on Zero to One (his new book) in that article and it dawned on me that is one of the intersting ways to look at the Data Scientist bubble.

Thiel's Classification of Innovation

Without trying to simplify and/or bastardize mr. Thiel's theory, the example in the Fortune Mag article will make this visible to most people (I hope). In the article the analogy is; going from one type writer to 100 type writers is 1 to N, inventing a word processor is moving us from 0 to 1. In other words, true innovation dramatically changes things by giving previously unknown power to the masses. It is that innovation that moves us from 0 to 1. Expansion of existing ideas - not true innovation - moves us from 1 to N. Of course, don't take my word on this but read the article or the book...

The Demise of the Human Data Scientist

The above paradigm explains the Data Scientist bubble quite nicely. Once upon a time companies hired a few PhD students who by chance had a degree in statistics and had learned how to program and figured out how to deal with (large) data sets. These newly minted data scientists proved that there is potential value in mashing data together, running analytics on these newly created data sets and thus caused a storm of publicity. Companies large and small are now frantically trying to hire these elusive data scientists, or something a little more down to earth, are creating data scientists (luckily not in the lab) by forming teams that bring a part of the skillset to the table.

This approach all starts to smell pretty much like a whole busload of typewriters being thrown at a well-known data analysis and data wrangling problem. Neither the problem nor the solution are new, nor innovative. Data Scientists are therefore not moving us from 0 to 1...

One could argue that while the data scientist quest is not innovative, at least is solves the problem of doing analytics. Fair and by some measure correct, but there is one bigger issue with the paradigm of "data scientists will solve our analytics problem" and that is scale. Giving the keys to all that big data to only a few data scientists is not going to work because these smart and amazing people are now becoming, often unbeknownst to them, an organizational bottleneck to gaining knowledge from big data.

The only real solution, our 0 to 1, is to expose a large number of consumers to all that big data, while enabling these consumers to apply a lot of the cool data science to all that data. In other words, we need to provide tools which include data science smarts. Those tools will enable us to apply the 80% common data science rules to the 80% of common business problems. This approach drives real business value at scale. With large chunks of issues resolved, we can then focus our few star data scientists on the 20% of problems or innovations that drive competitive advantage and change markets.

My Conclusion

The bubble is bursting because what I am seeing is more and more tools coming to market (soon) that will drive data science into the day-to-day job of all business people. Innovation is not the building of a better tool for data scientists or hiring more of them, instead the real 0 to 1 innovation is tools that make make all of us data scientists and lets us solve our own data science problems. The future of Data Science is smarter tools, not smarter humans.

Friday Sep 26, 2014

Why SQL is becoming the goto language for Big Data analysis

Since the term big data first appeared in our lexicon of IT and business technology it has been intrinsically linked to the no-SQL, or anything-but-SQL, movement. However, we are now seeing that SQL is experiencing a renaissance. The term “noSQL” has softened to a much more realistic approach "not-only-SQL" approach. And now there is an explosion of SQL-based implementations designed to support big data. Leveraging the Hadoop ecosystem, there is: Hive, Stinger, Impala, Shark, Presto and many more. Other NoSQL vendors such as Cassandra are also adopting flavors of SQL. Why is there a growing level of interest in the reemergence of SQL? Probably, a more pertinent question is: did SQL ever really go away? Proponents of SQL often cite the following explanations for the re-emergence of SQL for analysis:

  1. There are legions of developers who know SQL. Leveraging the SQL language allows those developers to be immediately productive.
  2. There are legions of tools and applications using SQL today.
  3. Any platform that provides SQL will be able to leverage the existing SQL ecosystem.

However, despite the virtues of these explanations, they alone do not explain the recent proliferation of SQL implementations. Consider this: how often does the open-source community embrace a technology just because it is the corporate orthodoxy? The answer is: probably not ever. If the open-source community believed that there was a better language for basic data analysis, they would be implementing it. Instead, a huge range of emerging projects, as mentioned earlier, have SQL at their heart The simple conclusion is that SQL has emerged as the de facto language for big data because, frankly, it is technically superior. Let’s examine the four key reasons for this:

  1. SQL is a natural language for data analysis.
  2. SQL is a productive language for writing queries.
  3. SQL queries can be optimised.
  4. SQL is extensible.

1. SQL is a natural language for data analysis.

The concept of SQL is underpinned by the relational algebra - a consistent framework for organizing and manipulating sets of data - and the SQL syntax concisely and intuitively expresses this mathematical system.

Most business users, data analysts and even data scientists think about data within the context of a spreadsheet. If you think about a spreadsheet containing a set of customer orders then what do most people do with that spreadsheet? Typically, they might filter the records to look only at the customer orders for a given region. Alternatively, they might hide some columns: maybe the customer address is not needed for a particular piece of analysis, but the customer name and their orders are important data points. Finally, they might add calculations to compute totals and/or perhaps create a cross tabular report.

Within the language of SQL these are common steps: 1) projections (SELECT), 2) filters and joins (WHERE), and 3) aggregations (GROUP BY). These are core operators in SQL. The vast majority of people have found the fundamental SQL query constructs to be straightforward and readable representation of everyday data analysis operations.

2. SQL is a productive language for writing queries.

When a developer writes a SQL query, he or she simply describes the results that they want. The developer does not have to get into any of the nitty-gritty of describing how to get the results 

This type of approach is often referred to as  'declarative programming,’ and it makes the developer's job easier. Even the simplest SQL query illustrates the benefits of declarative programming:

SELECT day, prcp, temp FROM weather
WHERE city = 'San Francisco' AND prcp > 0.0;

SQL engines may have multiple ways to execute this query (for example, by using an index). Fortunately the developer doesn't need to understand any of the underlying database processing techniques. The developer simply specifies the desired set of data using projections (SELECT) and filters (WHERE).

This is perhaps why SQL has emerged as such an attractive alternative to the MapReduce framework for analyzing HDFS data. MapReduce requires the developer to specify, at each step, how the underlying data is to be processed. For the same “query", the code is longer and more complex in MapReduce. For the vast majority of data analysis requirements, SQL is more than sufficient, and the additional expressiveness of MapReduce introduces complexity without providing significant benefits.

3. SQL queries can be optimized

The fact that SQL is a declarative language not only shields the developer from the complexities of the underlying query techniques, but also gives the underlying SQL engine has a lot of flexibility in how to optimize any given query. 

In a lot of programming languages, if the code runs slow, then it's the programmer's fault. For the SQL language, however, if a SQL query runs slow, then it's the SQL engine's fault.

This is where analytic databases really earn their keep – databases can easily innovate ‘under the covers’ to deliver faster performance; parallelization techniques, query transformations, indexing and join algorithms are just a few key areas of database innovation that drive query performance.

4. SQL is extensible

SQL provides a robust framework that adapts to new requirements

SQL has stayed relevant over the decades because, even though its core is grounded in universal data processing techniques, the language itself can be extended with new processing techniques and new calculations. Simple time-series calculations, statistical functions, and pattern-matching capabilities have all been added to SQL over the years. 

Consider, as a recent example, what many organizations realized as they started to ask queries such as 'how many distinct visitors came to my website last month?' These organizations realized that it is not vital to have a precise answer to this type of query ... an approximate answer (say, within 1%) would be more than sufficient. This has requirement has now been quickly delivered by implementing the existing hyperloglog algorithms within SQL engines for 'approximate count distinct' operations. 

More importantly, SQL is a language that is not explicitly tied to a storage model. While some might think of SQL as synonymous with relational databases, many of the new adopters of SQL are built on non-relational data. SQL is well on its way to being a standard language for accessing data stored in JSON and other serialized data structures.  


SQL is an immensely popular language today … and if anything its popularity is growing as the language is adopted for new data types and new use cases. The primacy of SQL for big data is not simply a default choice, but a conscious realization that SQL is the best suited language for basic analysis

PS. Next week, many sessions at this year’s OpenWorld will focus on the power, richness and performance of SQL for sophisticated data analysis including the following:

Monday September 28

Using Analytical SQL to Intelligently Explore Big Data @ 4:00PM Moscone North 131

Joerg Otto - Head of Database Engineering, IDS GmbH
Marty Gubar - Director, Oracle
Keith Laker - Senior Principal Product Manager, Data Warehousing and Big Data, Oracle

YesSQL! A Celebration of SQL and PL/SQL @ 6:00PM Moscone South 103

Steven Feuerstein - Architect, Oracle
Thomas Kyte - Architect, Oracle

Tuesday September 29

SQL Is the Best Development Language for Big Data @ 10:45AM Moscone South 104

Thomas Kyte - Architect, Oracle

Enjoy OpenWorld 2014 and if you have time please come and meet the Analytical SQL team in the Moscone South Exhbition Hall. We will be on the Parallel Execution and Advanced SQL Processing demo booth (id 3720).

Tuesday Sep 23, 2014

Big Data IM Reference Architecture

Just in time for Oracle Openworld, the new Big Data Information Management Reference Architecture is posted on our OTN pages. The reference architecture attempts to create order in the wild west of new technologies, the flurry of new ideas and most importantly tries to go from marketing hype to a real, implementable architecture.

To get all the details, read the paper here. Thanks to the EMEA architecture team , the folks at Rittman Mead Consulting and all others involved.

Tuesday Jul 22, 2014

StubHub Taps into Big Data for Insight into Millions of Customers’ Ticket-Buying Patterns, Fraud Detection, and Optimized Ticket Prices

The benefits of Big Data at Stubhub:

  • Stubhub enabled data scientists to work directly with customer-related data—such as ticket-purchasing history—inside the database, and to use database options to explore the data graphically, build and evaluate multiple data-mining models, and deploy predictions and insights throughout the enterprise—drastically improving StubHub’s agility and responsiveness
  • Developed highly targeted ticket promotional campaigns and offers by having the ability to calculate 180 million customers’ lifetime value (or propensity) instead of just 20,000 values at a time
  • Used Oracle R Enterprise component of Oracle Advanced Analytics—an Oracle Database option—to reduce a fraud issue by up to 90%

Read more or watch the video:

Thursday Jun 26, 2014

Big Data Breakthrough; Watch the Webcast on July 15th

Wednesday Jun 25, 2014

Big Data Appliance now part of Exastack Partner Program

With data growing at unprecedented levels, application providers need a robust, scalable and high-performance infrastructure to manage and analyze data at the speed of business. Helping Independent Software Vendors (ISVs) meet these challenges, Oracle PartnerNetwork (OPN) is today extending its Oracle Exastack program to give partners the opportunity to achieve Oracle Exastack Ready or Optimized status for Oracle Big Data Appliance and Oracle Database Appliance.

The announcement was made today during the sixth annual Oracle PartnerNetwork (OPN) global kickoff event, “Engineered for Success: Oracle and Partners Winning Together.” For a replay of the event, visit:

Here is a word from one of our partners: “TCS was looking for an enterprise grade and reliable Hadoop solution for our HDMS document management application,” said Jayant Dani, principal consultant, Component Engineering Group, TCS.  “We were drawn to Oracle's Big Data Appliance for its ready to use platform, superior performance, scalability, and simple scale-out capabilities.”

Read the entire news release here.

Thursday Jun 05, 2014

Globacom and mCentric Deploy BDA and NoSQL Database to analyze network traffic 40x faster

In a fast evolving market, speed is of the essence. mCentric and Globacom leveraged Big Data Appliance, Oracle NoSQL Database to save over 35,000 Call-Processing minutes daily and analyze network traffic 40x faster. 

Here are some highlights from the profile:

Why Oracle

“Oracle Big Data Appliance works well for very large amounts of structured and unstructured data. It is the most agile events-storage system for our collect-it-now and analyze-it-later set of business requirements. Moreover, choosing a prebuilt solution drastically reduced implementation time. We got the big data benefits without needing to assemble and tune a custom-built system, and without the hidden costs required to maintain a large number of servers in our data center. A single support license covers both the hardware and the integrated software, and we have one central point of contact for support,” said Sanjib Roy, CTO, Globacom.

Implementation Process

It took only five days for Oracle partner mCentric to deploy Oracle Big Data Appliance, perform the software install and configuration, certification, and resiliency testing. The entire process—from site planning to phase-I, go-live—was executed in just over ten weeks, well ahead of the four months allocated to complete the project.

Oracle partner mCentric leveraged Oracle Advanced Customer Support Services’ implementation methodology to ensure configurations are tailored for peak performance, all patches are applied, and software and communications are consistently tested using proven methodologies and best practices.

Read the entire profile here.

Tuesday Apr 22, 2014

Announcing: Big Data Appliance 3.0 and Big Data Connectors 3.0

Today we are releasing Big Data Appliance 3.0 (which includes the just released Oracle NoSQL Database 3.0) and Big Data Connectors 3.0.These releases deliver a large number of interesting and cool features and enhance the overall Oracle Big Data Management System that we think is going to be the core of information management going forward.

This post highlights a few of the new enhancements across the BDA, NoSQL DB and BDC stack.

Big Data Appliance 3.0:
  • Pre-configured and pre-installed CDH 5.0 with default support for YARN and MR2
  • Upgrade from BDA 2.5 (CDH 4.6) to BDA 3.0 (CDH 5.0)
  • Full encryption (at rest and over the network) from a single vendor in an appliance
  • Kerberos and Apache Sentry pre-configured
  • Partition Pruning through Oracle SQL Connector for Hadoop
  • Apache Spark (incl. Spark Streaming) support
  • More
Oracle NoSQL Database 3.0:
  • Table data model support layered on top of distributed key-value model
  • Support for Secondary Indexing
  • Support for "Data Centers" => Metro zones for DR and secondary zones for read-only workloads
  • Authentication and network encryption
  • More

You can read about all of these features by going to links above and reading the OTN page, data sheets and other relevant information.

While BDA 3.0 immediately delivers upgrade from BDA 2.5, Oracle will also support the current version and we fully expect more BDA 2.x releases based on more CDH 4.x releases. As a customer you now have a choice how to deploy BDA and which version it is you want to run, while knowing you can upgrade to the latest and greatest in a safe manner.

Thursday Apr 03, 2014

Updated: Price Comparison for Big Data Appliance and Hadoop

Untitled Document

It was time to update this post a little. Big Data Appliance grew, got more features and prices as well as insights just changed all across the board. So, here is an update.

The post is still aimed at providing a simple apples-to-apples comparison and a clarification of what is, and what is not included in the pricing and packaging of Oracle Big Data Appliance when compared to "I'm doing this myself - DIY style".

Oracle Big Data Appliance Details

A few of the most overlooked items in pricing out a Hadoop cluster are the cost of software, the cost of actual production-ready hardware and the required networking equipment. A Hadoop cluster needs more than just CPUs and disks... For Oracle Big Data Appliance we assume that you would want to run this system as a production system (with hot-pluggable components and redundant components in your system). We also assume you want the leading Hadoop distribution plus support for that software. You'd want to look at securing the cluster and possibly encrypting data at rest and over the network. Speaking of network, InfiniBand will eliminate network saturation issues - which is important for your Hadoop cluster.

With that in mind, Oracle Big Data Appliance is an engineered system built for production clusters.  It is pre-installed and pre-configured with Cloudera CDH and all (I emphasize all!) options included and we (with the help of Cloudera of course) have done the tuning of the system for you. On top of that, the price of the hardware (US$ 525,000 for a full rack system - more configs and smaller sizes => read more) includes the cost of Cloudera CDH, its options and Cloudera Manager (for the life of the machine - so not a subscription).

So, for US$ 525,000 you get the following:

  • Big Data Appliance Hardware (comes with Automatic Service Request upon component failures)
  • Cloudera CDH and Cloudera Manager
  • All Cloudera options as well as Accumulo and Spark (CDH 5.0)
  • Oracle Linux and the Oracle JDK
  • Oracle Distribution of R
  • Oracle NoSQL Database Community Edition
  • Oracle Big Data Appliance Enterprise Manager Plug-In

The support cost for the above is a single line item.. The list price for Premier Support for Systems per the Oracle Price list (see source below) is US$ 63,000 per year.

To do a simple 3 year comparison with other systems, the following table shows the details and the totals for Oracle Big Data Appliance. Note that the only additional item is the install and configuration cost which are done by Oracle personnel or partners, on-site:

Year 1 Year 2 Year 3 3 Year
BDA Cost

Annual Support Cost

On-site Install (approximately)


For this you will get a full rack BDA (18 Sun X4-2L servers, 288 cores (Two Intel Xeon E5-2650V2 CPUs per node), 864TB disk (twelve 4TB disks per node), plus software, plus support, plus on-site setup and configuration. Or in terms of cost per raw TB at purchase and at list pricing: $697.

HP DL-380 Comparative System (this is changed from the original post to the more common DL-380's)

To build a comparative hardware solution to the Big Data Appliance we picked an HP-DL180 configuration and built up the servers using the website for pricing. The following is the price for a single server.

Model Number Description Quantity Total Price
653200-B21 ProLiant DL380p Gen8 Rackmount Factory Integrated 8 SFF CTO Model (2U) with no processor, 24 DIMM with no memory, open bay (diskless) with 8 SFF drive cage, Smart Array P420i controller with Zero Memory, 3 x PCIe 3.0 slots, 1 FlexibleLOM connector, no power supply, 4 x redundant fans, Integrated HP iLO Management Engine
2.6GHz Xeon E5-2650 v2 processor (1 chip, 8 cores) with 20MB L3 cache - Factory Integrated Only
HP 1GbE 4-port 331FLR Adapter - Factory Integrated Only
460W Common Slot Gold Hot Plug Power Supply
HP Rack 10000 G2 Series - 10842 (42U) 800mm Wide Cabinet - Pallet Universal Rack
8GB (1 x 8GB) Single Rank x8 PC3L-12800R (DDR3-1600) Registered CAS-11 Low Voltage Memory Kit
HP Smart Array P222/512MB FBWC 6Gb 1-port Int/1-port Ext SAS controller 1
4TB 6Gb SAS 7.2K LFF hot-plug SmartDrive SC Midline disk drive (3.5") with 1-year warranty

Grand Total for a single server (list prices)


On top of this we need InfiniBand switches. Oracle Big Data Appliance comes with 3 IB switches, allowing us to expand the cluster without suddenly requiring extra switches. And, we do expect these machines to be a part of a much larger clusters. The IB switches are somewhere in the neighborhood of US$ 6,000 per switch, so add $18,000 per rack and add a management switch (BDA uses a Cisco switch) which seems to be around $15,000 list. The total switching comes to roughly $33,000.

We will also need Cloudera Enterprise subscription - and to compare apples to apples, we will do it for all software. Some sources (see this document) peg CDH Core at $3,382 list per node and per year (24*7 support). Since BDA has more software (all options) and that pricing is not public I am going to make an educated calculation and rounding and double the price with a rounding to the nearest nice and round number. That gets me to $7,000 per node, per year for 24*7 support. 

BDA also comes with on-disk encryption, which is even harder to price out. My somewhat educated guess is around $1,500 list or so per node and per year. Oh, and lets not forget the Linux subscription, which lists at $1,299 per node per year. We also run a MySQL database (enterprise edition with replication), which costs list subscription $5,000. We run it replicated over 2 nodes.

This all gets us to roughly $10,000 list price per node per year for all applicable software subscriptions and support and an additional $10,000 for the two MySQL nodes.

HP + Cloudera Do-it-Yourself System

Let's go build our own system. The specs are like a BDA, so we will have 18 servers and all other components included. 

Year 1 Year 2 Year 3 Total




SW Subscriptions and Support

Installation and Configuration


Some will argue that the installation and configuration is free (you already pay your data center team), but I would argue that something that takes a short amount of time when done by Oracle, is worth the equivalent if it takes you a lot longer to get all this installed, optimized, and running. Nevertheless, here is some math on how to get to that cost anyways: approximately 150 hours of labor per rack for the pure install work. That adds up to US $15,000 if we assume a cost per hour of $100. 

Note: those $15,000 do NOT include optimizations and tuning to Hadoop, to the OS, to Java and other interesting things like networking settings across all these areas. You will now need to spend time to figure out the number of slots you allocate per node, the file system block size (do you use Apache defaults, or Cloudera's or something else) and many more things at system level. On top of that, we pre-configure for example Kerberos and Apache Sentry giving you a secure authorization and authentication method, as well as have a one-click on-disk and network encryption setting. Of course you can contact various other companies to do this for you.

You can also argue that "you want the cheapest hardware possible", because Hadoop is built to deal with failures, so it is OK for things to regularly fail. Yes, Hadoop does deal well with hardware failures, but your data center is probably much less keen about this idea, because someone is going to replace the disks (all the time). So make sure the disks are hot-swappable. An oh, that someone swapping the disks does cost money... The other consideration is failures in important components like power... redundant power in a rack is a good thing to have. All of this is included (and thought about) in Oracle Big Data Appliance.

In other words, do you really want spend weeks installing, configuring and learning or would you rather start to build applications on top of the Hadoop cluster and thus providing value to your organization.

The Differences

The main differences between Oracle Big Data Appliance and a DIY approach are:

  1. A DIY system - at list price with basic installation but no optimization - is a staggering $220 cheaper as an initial purchase
  2. A DIY system - at list price with basic installation but no optimization - is almost $250,000 more expensive over 3 years.
    Note to purchasing, you can spend this on building or buying applications on your cluster (or buy some real intriguing Oracle software)
  3. The support for the DIY system includes five (5) vendors. Your hardware support vendor, the OS vendor, your Hadoop vendor, your encryption vendor as well as your database vendor. Oracle Big Data Appliance is supported end-to-end by a single vendor: Oracle
  4. Time to value. While we trust that your IT staff will get the DIY system up and running, the Oracle system allows for a much faster "loading dock to loading data" time. Typically a few days instead of a few weeks (or even months)
  5. Oracle Big Data Appliance is tuned and configured to take advantage of the software stack, the CPUs and InfiniBand network it runs on
  6. Any issue we, you or any other BDA customer finds in the system is fixed for all customers. You do not have a unique configuration, with unique issues on top of the generic issues.


In an apples-to-apples comparison of a production Hadoop cluster, Oracle Big Data Appliance starts of with the same acquisition prices and comes out ahead in terms of TCO over 3 years. It allows an organization to enter the Hadoop world with a production-grade system in a very short time reducing both risk as well as reducing time to market.

As always, when in doubt, simply contact your friendly Oracle representative for questions, support and detailed quotes.


HP and related pricing: or (the latter is a paid service - sorry!)
Oracle Pricing:
MySQL Pricing:

Wednesday Mar 19, 2014

Announcing Encryption of Data-at-Rest on Big Data Appliance

With the release of Big Data Appliance software bundle 2.5, BDA completes the encryption story underneath Cloudera CDH. BDA already came with network encryption, ensuring no network sniffing can be applied in between the nodes, it now adds encryption of data-at-rest.

A Brief Overview

Encryption of data-at-rest can be done in 2 modes. One mode leverages the Trusted Platform Module (TPM) on the motherboard to provide a key to encrypt the data on disk. This mode does not require a password or pass phrase but relies on the motherboard. The second mode leverages a passphrase, which in turn will be used to generate a private-public key pair generated with OpenSSL. The key pair is encrypted as well.

The passphrase encryption has a few more interesting aspects. For one, it does require the passphrase to be entered upon re-booting the system. Leveraging the TPM option does not require any manual intervention at reboot. On Big Data Appliance it is possible to regularly change the passphrase without impacting the encryption, or required re-encryption of the data.

Neither one of the encryption methods affect user access to user data. In other words, on an unprotected cluster a user that can read data before encryption will be able to read data after encryption. The goal is to ensure data is protected on physical media - like theft or incorrect disposal of a disk. Both forms protect from that, but only passphrase based encryption protects from disposal or theft of a server.

On BDA, it is possible to switch between these two methods. This does have impact on running the cluster as data needs to be re-encrypted. For this step the cluster will be down, however data is not duplicated, so there is no need to reserve double the space to do the re-encryption.

How to Encrypt Data

As with all installation or changes on Big Data Appliance you will leverage Mammoth to do the install with encryption or to make changes to the system if you are already in production. Before you set up either of the two modes of data-at-rest encryption, you should consider your requirements. Changing the mode - as described - is possible, but will require the cluster to be down for re-encryption.

Full Set of Security Features

Encryption - out-of-the-box is yet another feature that is specific to Oracle Big Data Appliance. On top of pre-configured Kerberos, Apache Sentry, Oracle Audit Vault Encryption now adds another security dimension. To read more about the full set of features start here.

Thursday Mar 13, 2014

Video: Big Data Connectors and IDH (Strata)

The certification of Oracle Big Data Connectors on Intel Distribution for Hadoop now complete (see our previous post). This video from Strata gives you a nice overview of IDH and BDC.


The data warehouse insider is written by the Oracle product management team and sheds lights on all thing data warehousing and big data.


« May 2016