Data Stewardship in 13 Minutes a Week

February 14, 2013

Andrew White, Research VP at Gartner, has a great thesis on data stewardship:

“13 minutes a week – that is how much work your data stewards should be doing.”

That is, in order for data stewardship (or data ownership) to be truly adopted by the business—marketing, HR, finance—the work we require them to do should amount to no more than 13 minutes per week.

This is a terrific goal and it is what we strive for withDataPrivilege.  How do we do it?

  • We make reviews devoid of noise – stewards only see data they care about
  • We make reviews differential – if it hasn’t changed since last review, it doesn’t show up
  • We make reviews inline with normal workflow – a timely email appears in the steward’s inbox with a big link that takes them right to the review; no separate reminders or TODOs needed
  • We make reviews actionable – exceptional items are highlighted and a suggested action is given along with the ability to take the action without leaving the review screen

A significant portion of our operational plan is devoted to finding, assigning, and involving data owners.  But without buy-in from the people who will be doing the work, the plan can’t be executed.  Andrew cuts right to the core of why many businesses have failed at implementing information governance programs: they have effectively dumped an unreasonable and unnecessaryamount of work on their stewards’ desks and walked away.

What do you think? Could you sell 13 minutes of work per week in exchange for true information governance, accountability, and data protection?


Hello, Hekaton! Microsoft Plans In-Memory OLTP SQL Server

November 28, 2012

In a keynote at SQL PASS Summit, Microsoft announced it is bringing In-Memory online transaction processing (OLTP) to the next major release of SQL Server, code named Hekaton. Twitter lit up with chatter about it, and at our booth at PASS, Hekaton was the topic du jour.

Our quick take on Hekaton is that even in-memory databases have transaction logs that are on persistent storage. By utilizing ioMemory as the persistent storage for transaction logs, we accelerate transaction processing, as the logs are being written to constantly. It’s faster to recover an in-memory database as well, because the transaction logs have to be read quickly to recover a database fast. Finally, the backup speed of in-memory databases gets faster if the backups reside on ioMemory.

Hekaton is all about efficiency and performance, the same ideals that drive innovation at Fusion-io. We think Hekaton sounds promising. Here’s to database acceleration!

Here are a few tidbits about Hekaton from around the web:

Hekaton is the Greek word for 100 times, and Microsoft says that’s the design goal for the peak performance improvements it’s expecting.” Doug Henschen, Informationweek

Hekaton is currently in private technology preview with a small set of customers, which company officials are planning to expand to 100 before the end of this calendar year.” Mary Jo Foley, ZDNet

The next version of SQL Server will feature the ability to host database tables or even entire databases within a server’s working memory.” Joab Jackson, Computerworld

Data has emerged as the new currency of business.” Ted Kummert, Corporate Vice President of Microsoft’s Business Platform Division, The Official Microsoft Blog

In a recent article in FX-MM, Steven Graves, CEO of McObject, reports on his experience with ioMemory and in-memory databases. Check out “Databases: Have your in-memory performance, and recoverability too.”

 


The Formula for Analytics Success: Data Knowledge

November 12, 2012

Companies spend a small fortune continually investing and reinvesting in making their business analysts self-sufficient with thelatest and greatest analytical tools. Most companies have multiple project teams focused on delivering tools to simplify and improve business decision making. There are likely several standard tools deployed to support the various data analysis functions required across the enterprise: canned/batch reports, desktop ad hoc data analysis, and advanced analytics. There’s never a shortage of new and improved tools that guarantee simplified data exploration, quick response time, and greater data visualization options, Projects inevitably include the creation of dozens of prebuilt screens along with a training workshop to ensure that the users understand all of the new whiz bang features associated with the latest analytic tool incarnation.  Unfortunately, the biggest challenge within any project isn’t getting users to master the various analytical functions; it’s ensuring the users understand the underlying data they’re analyzing.

If you take a look at the most prevalent issue with the adoption of a new business analysis tool is the users’ knowledge of the underlying data.  This issue becomes visible with a number of common problems:  the misuse of report data, the misunderstanding of business terminology, and/or the exaggeration of inaccurate data.  Once the credibility or usability of the data comes under scrutiny, the project typically goes into “red alert” and requires immediate attention. If ignored, the business tool quickly becomes shelfware because no one is willing to take a chance on making business decisions based on risky information.

All too often the focus on end user training is tool training, not data training. What typically happens is that an analyst is introduced to the company’s standard analytics tool through a “drink from a fire hose” training workshop.  All of the examples use generic sales or HR data to illustrate the tool’s strengths in folding, spindling, and manipulating the data.  And this is where the problem begins:  the vendor’s workshop data is perfect.  There’s no missing or inaccurate data and all of the data is clearly labeled and defined; classes run smoothly, but it just isn’t reality  Somehow the person with no hands-on data experience is supposed to figure out how to use their own (imperfect) data. It’s like someone taking their first ski lesson on a cleanly groomed beginner hill and then taking them up to the top of an a black diamond (advanced) run with step hills and moguls.  The person works hard but isn’t equipped to deal with the challenges of the real world.  So, they give up on the tool and tell others that the solution isn’t usable.

 

All of the advanced tools and manipulation capabilities don’t do any good if the users don’t understand the data. There are lots of approaches to educating users on data.  Some prefer to take a bottom-up approach (reviewing individual table and column names, meanings, and values) while others want to take a top-down approach (reviewing subject area details, the associated reports, and then getting into the data details).  There are certainly benefits of one approach over the other (depending on your audience); however, it’s important not to lose sight of the ultimate goal: giving the users the fundamental data knowledge they need to make decisions.  The fundamentals that most users need to understand their data include a review of

The above details may seem a bit overwhelming if you consider that most companies have mature reporting environments and multi-terabyte data warehouses.  However, we’re not talking about training someone to be an expert on 1000 data attributes contained within your data warehouse; we’re talking about ensuring someone’s ability to use an initial set of reports or a new tool without requiring 1-on-1 training.  It’s important to realize that the folks with the greatest need for support and data knowledge are the newbies, not the experienced folks.

There are lots of options for imparting data knowledge to business users:  a hands-on data workshop, a set of screen videos showing data usage examples, or a simple set of web pages containing definitions, textual descriptions, and screen shots. Don’t get wrapped up in the complexities of creating the perfect solution – keep it simple.  I worked with a client that deployed their information using a set of pages constructed with PowerPoint that folks could reference in a the company’s intranet. If your users have nothing – don’t’ worry about the perfect solution – give them something to start with that’s easy to use.

Remember that the goal is to build users’ data knowledge that is sufficient to get them to adopt and use the company’s analysis tools.  We’re not attempting to convert everyone into data scientists; we just want them to use the tools without requiring 1-on-1 training to explain every report or data element.

Thanks to http://evanjlevy.wordpress.com/2012/11/12/the-formula-for-analytics-success-data-knowledge/


Rhapsody Tunes Performance Scaling

October 2, 2012

Rhapsody’s digital music service gives subscribers unlimited on-demand access to more than 16 million songs and promises best-in-class service no matter what device its customers use. As Rhapsody’s customer base has expanded and its database of songs has grown, the demand on its supporting IT infrastructure increased to the point where latency issues became a problem.

Rhapsody’s VP of Platform and Operations Heng Cao solved these challenges with Fusion’s ioMemory. After implementing ioMemory on its Oracle database content management system, batch processing job times ran ten times faster.

Seeing this success, Heng added ioMemory to Rhapsody’s transactional database, which sped job times 8X.

Heng even found he could meet his High Availability needs with Oracle Data Guard. “With the ioDrive2s, we now have the capacity to host Oracle Data Guard replicated databases within the servers, while still leaving plenty of room for future growth,” he told us.

But best of all, the cost savings of the Fusion Powered solution paid for itself. Heng said, “We realized 100% ROI just on the cost of the Oracle database licenses we would have needed had we implemented a storage-based system.”

Heng is thrilled with his Fusion Powered systems and is looking to implement ioDrive2s in other I/O-constrained systems. “Based on success in production, we are currently deploying ioMemory into our preproduction environment to boost development and testing productivity, and are seeing great success.”


Extreme Performance, Lower Cost: HP’s New Data Accelerator Solution for Oracle Database

September 20, 2012

We first talked about the HP Data Accelerator Solution for Oracle Database in December 2011 in our blog. Recently, HP revised this solution and whitepaper with an option for high availability using Oracle Data Guard, enabling the creation of a multi-server solution which guards against hardware or database failures by eliminating single points of failure. This is HP’s Recommended Architecture for OLTP with Business Continuity and is based on the HP ProLiant DL980 server configured with Fusion-powered HP IO Accelerators as the keystone of the solution.

The HP Data Accelerator Solution for Oracle Database delivers extreme performance for OLTP database workloads without an extreme price tag, without high operating costs, and without locking you or your IT budget into an all-Oracle server solution. HP starts with the enterprise-class HP ProLiant DL980 server and replaces traditional external mechanical storage with HP IO Accelerators—PCIe-based high-capacity, enterprise-quality flash-memory modules. Using IO Accelerators as a new in-server storage memory tier moves terabytes of process-critical data closer to the server processor to dramatically improve performance. You can accomplish this performance boost by either offloading the database’s most active data sets or hosting the entire database on the IO Accelerators, which have microsecond data access latencies—orders of magnitude lower latency than traditional storage. The result is a database solution which can deliver more than four times the transaction throughput of a competitive Oracle Exadata solution at half the acquisition cost and one-quarter the operational cost, according to the whitepaper.

Note that all data reported in this paper is based on HP’s testing performed with Gen1 IO Accelerators and completed prior to the availability of HP’s current, Gen2 IO Accelerators, which provide roughly double the capacity and performance. So while this solution with Gen1 IO Accelerators does offer compelling advantages, use of the Gen2 IO Accelerators in this solution can deliver significantly higher transaction throughput, better cost economics and other benefits.

By deploying IO Accelerators in the server, the solution configuration is very compact – it requires 70% less datacenter space than comparable traditional server configurations. It’s also substantially simpler and more open than other solutions, including the rigid and inflexible Exadata. HP built this solution, with its capacity for multi-terabyte databases, from the ground up for OLTP workloads, rather than re-treading a data warehouse design. With IO Accelerators, the maximum capacity for the HP DL980 server is 25.2 terabytes, (six 2.4TB and nine 1.2TB IO Accelerators).

Get the full HP Whitepaper to learn more about this new HP Data Accelerator Solution for Oracle Database.

For more information about what fusion-io can do for your business applications please contact C24 at http://www.c24.co.uk

 


COMPETING ON ANALYTICS: AN ARTICLE REVIEW

July 9, 2012

A Harvard Business Review Article by Thomas H. Davenport, Article Review by Akhmad Rahadian Hutomo

Since the late of 1990s, the term business intelligence (BI) and its application has been widely known and used in organizations, especially in large enterprises. But in a decade later, they started to realize that changing business environment will needs something more than just BI, which now called business analytics. In 2006, an author named Thomas wrote an article on HBR entitled “Competing on Analytics” which provisions the rising needs for business analytics. Davenport started his explanation on competing analytics by giving some examples on the succesfull usage of killer apps in some organizations, named Amazon, Harrah’s, Capital One and Boston Red Sox. By utilizing analytics, these organizations are able to knows better about the values that customer want, which inturn be able to squeeze all the value from the processes and make the best out of it. Davenport also point out that, to be an analytics competitor, top-down approach from the senior leadership team, as well as hiring the best people are necessary. Nonetheless, not all organizations are succesfull on using business analytics due to its characteristic. The rest of the articles explains about what organizations can make the best of analytics, as well as the changes that an organization must undergo to adopt it.

ANATOMY OF AN ANALYTICS COMPETITOR: MUST-HAVE CHARACTERISTICS FOR ORGANIZATIONS

Some traditional organizations may not be fully suitable with competing analytics. One best practice that an organization my want to know is how Marriot International using analytics. But, it will not work to some traditional organizations. Davenport’s study found three key attributes that an organization must have:

WIDESPREAD USE OF MODELLING AND OPTIMIZATION

Analytics competitors do things beyond statistics and spreadsheets. They are using sort of things that could provide them better insights from data, such as:

  • Predictive modelling to identify the most profitable customer.
  • Data warehouse to pool inhous and outside data.
  • Optimized supply chain.
  • Real-time pricing.
  • Sophisticated experiments to calculate impact.

Some analytics competitors, especially inscurance company, like Capital One and Progessive doing series comprehensive experiments to have the best value based on their customers need, even with high-risk.

AN ENTERPRISE APPROACH

Successfull analytics competitor will implement analytics using multiple applications in wide busines functions rather than using single app. For some companies such as UPS, Capital One and Barclays Bank are already implementing business intelligence and then shifting towards full-bore analytics competitors. However, Devenport thinks that BI still have some flaws where its still use data which spreads all over the organization. The data may contains errors and make the decision inacurrate, which in contrast, analytics competitors are using centralized function to manage critical data. People within the organization is as important as the technology. Some organization like P&G create a pool of experts from various function to do the analytics.

SENIOR EXCECUTIVE ADVOCATES

Changing into an analytics competitor simply changes the organization, and it will require leadership skills to guide the organization towards sucessfull adoption. Its proven that if the initiative just pushed by one-or-two business unit leaders, it will not successfull. There was some key leadership qualities that the article pointed out, such as: appreciation and familiarity with analytics or analytics-minded, intuitive, and have the guts to make decision even not supported by numbers.

THEIR SOURCES OF STRENGTH: WHAT MAKES AN ANALYTICS COMPETITOR RUNS

Basically Davenport define 4 things that makes an analytics competitor ticks, they are:

THE RIGHT FOCUS: HAVING A CLEAR SIGHT

Even if an organization have the ability, it is necessary to have certain focus on only a few analytics subjects. Becoming to diffuse can make the organization losing clear sight on the purpose of analytics. Another consideration of focus is about having a deep analysis on at least 7 functions. Nowadays, advanced statistic models and algorithm ca be used widely, including in advertising and other marketing measures. Later on this subtopic, there are examples that sucessfull analytics competitors can’t be done by the organization alone, it also needs to help their vendors and customers.

THE RIGHT CULTURE: TO JUSTIFY EVERYTHING QUICKLY

The right culture to have is the culture to appreciate usage of data, fact and the things between that and the procedure to get it. It also applied in organization with high creativity and intrapreneurship: any innovation should be made based on evidence. However, always justify everything also have payoff: it might be taking long time and costly, so the managers hould balance them in order to make quick decisions.

THE RIGHT PEOPLE: THE BEST OF THEM

Analytics competitors hires best people on analytics, bunch of them, to do the analytic-based decisions and make it seamlessly in line with the business. But, the people to do the analytics just as good as how far they can communicate it, so they must have sort of good interpersonal skills. In terms of formula, it might look like this:

Good Analyst = Expertise +Ablity to express it in simple way + Interpersonal skills

Of course, to get people with this quality is not easy, not to mention taking long waiting time. To have an overseas employee might be a good idea.

THE RIGHT TECHNOLOGY: THREE PILLARS

Analytics and IT are unseparable. It is supported by three pillars: First, THE DATA,whether it is from ERP, CRM, POS, any of them, and a lot of them, means years of data. They put it in data warehouse, which a familiar tools on BI. Second, THE BI SOFTWARE, to collect data from warehouses, analyse them and making reports. And Last, THE COMPUTING HARDWARE which enables a computation power for huge volume of data, quickly.

THE (LONG) ROAD AHEAD

Well, it might be not long road, as Davenport writhe the articles 5 years before this review written in the late 2011. He was concluding his paper with reminding us that to become an analytics competitor will takes a long time until the ROI, while meantime, it will cost many efforts and expenses. Yet, it can be done gradually from current time by collecting data and refining the system, and equip the organization with analytics-minded people.

COMMENTARY

Business analytics might be an interestring concept to explore to enrich our current knowledge and view on today’s business intelligence. In contrast with BI, business analytics focuses on gaining insights and overview of organizational performance based on data and statistical methods, supported by BI applications. It also cover the issues of leadership, culture and having a certain quality of analyst within the organization. On the article, Davenport gives the readers a comprehensive look of business analytics without losing the big picture. His writing also well supported with examples which gives personal and easy-to-digest touch on complex concept. A worth to read for BI enthusiasts.

Based on a Harvard Business Review Article Titled “Competing on Analytics” by Thomas H. Davenport Published on January 2006, Article Review By Akhmad Rahadian Hutomo for Business Intelligence Assignment, Information System, Faculty of Computer Science, Universitas Indonesia on October 2011.

http://ianhutomo.wordpress.com/2012/07/07/business-intelligence-in-human-capital-driven-companies/

 


Integrate Data into Products, or Get Left Behind

July 9, 2012

Over the last several years, interest in and excitement about analytics/big data/data mining has grown exponentially. Count me among its biggest enthusiasts, as I firmly believe the potential for the “this changes everything” discoveries are real!

I’m just as excited about “informationalization,” a concept that’s been around for a whilebut has been gaining speed in recent years. The basic idea is simple: Make existing products and services more valuable to your customers by building in more data and information.

Read More


BrainPad Accelerates Multiple Web Analytics Systems on Less Hardware : fusion-io

July 6, 2012

BrainPad, Inc. provides Web-based data mining, business analytics, operational research, and mathematical solutions for businesses. Its L2Mixer service provides business intelligence on end-user pay-per-click behavior, allowing companies to optimize product pricing. BrainPad’s Rtoaster service provides end-users product recommendations based on a behavioral analysis of their browsing patterns.

As both customer base and product line expanded, so did the load on both systems’ databases, which threatened to slow performance. BrainPad needed to increase processing speeds of its L2Mixer solution, increase analysis times of its Rtoaster system, and do both with reduced costs and maintenance. Tsuyoshi Inoue, BrainPad’s Chief Engineering Architect, was impressed with the way Fusion ioDrives resolved all three problems.

Tsuyoshi said, “One batch job [on the PostgreSQL system], which used to take over four hours to complete, ran in less than 30 minutes. The ioDrives also doubled the number of threads we could run in parallel. Performance is high enough that we can now meet the most demanding customer SLAs.”

Shifting the L2Mixer databases from hard disks to the ioDrives cut I/O wait time by more than half, resulting in 29 time faster aggregate data calculation and 10 times faster summary data reports. Moving the Rtoaster’s PostgreSQL database from disks to ioDrives sped batch job processing by 30 times.

“Before adding Fusion-io,” Tsuyoshi explained, “we had to run database maintenance tasks once a week or more just to avoid a serious performance degradation. Now, we can eliminate these tasks altogether, which is quite significant. Our new system is more simpler, more flexible, and easier to modify and improve.”

Want to see more astounding results BrainPad achieved with its Fusion Powered system? Read the BrainPad case study.


1.9M row insertions​/sec — HP IO Accelerato​rs, HP ProLiant Gen8 servers & Microsoft SQL Server 2012

June 7, 2012

In terms of passenger throughput, Shinjuku Station in Japan is considered the world’s busiest train station. Connecting rail traffic between central Tokyo and its western suburbs, Shinjuku recorded an average 3.64 million passengers per day in 2007. Now imagine upping that to 10 million per day without adding any new trains and the associated cost. That’s sort of what it’s like to reach 1.9 million row insertions per second in a database table using the new HP ioDrive2 IO Accelerators configured within HP ProLiant Gen8 servers.

If you work with databases, you know that 1.9 million row insertions per second is much more than a lot. We knew that the new generation of HP IO Accelerators, released on May 14th together with the next wave of HP ProLiant Gen8 servers, would provide our customers with a tremendous application performance boost. And recent testing in a Microsoft SQL Server 2012 environment made us virtually giddy.

HP IO Accelerators place data close to the server CPU, dramatically improving application performance by bypassing traditional storage controllers and unlocking trapped compute cycles. Based on ioMemory technology from Fusion-io, they operate as a new storage memory tier within the server.  These devices install on the PCIe bus within your rack, tower, or blade server to help eliminate I/O bottlenecks, performance limitations, latency problems, and database blocking.  For database applications such as Microsoft SQL 2012, you can accomplish this by (1) hosting the entire database in the server (and mirror with a second server if you need high availability), or by (2) using HP IO Accelerators as a high performance read cache for frequently accessed files (such as log files), with the balance of the dataset hosted by external storage.

Three Microsoft SQL Server 2012 database transaction log stress tests were conducted over the past seven months with the 2nd generation HP IO Accelerators. Earlier this month, we created a Microsoft SQL Server environment using a single HP ProLiant DL380p Gen8 server configured with two of the new 1.2 TB HP ioDrive2 IO Accelerators. Comparing this testing  with results from the same test operations performed with two other configurations, tested in November 2011 and March 2012, we saw a huge lift in performance.  The November test used four ioDrive2 devices with a single 4-socket server.  The March test was conducted live at SQLBits in London using three ioDrive2 devices with a single HP DL380 G7 server. This test was structured in particular to maximize use of the transaction log, and what we found was stunning:  The environment with the HP DL380p Gen8 server and two HP ioDrive2 IO Accelerators easily sustained over 900 MB/s of log file access – 72% higher performance than the November 2011 tests and 20% more than the March 2012 tests.  And this environment provided 1.9 million row insertions per second – representing 50% more bandwidth than the November 2011 tests and 20% more than the March 2012 tests.  And, this was accomplished by using only two HP ioDrive2 IO Accelerators within a single HP DL380p Gen8 server.  Better performance with fewer cards – that’s a crowd pleaser!

Dynamic workload acceleration is a huge part of the Gen8 story. By pairing HP IO Accelerators with HP ProLiant Gen8 servers and HP Insight Control management software, you have an ideal platform for Microsoft SQL Server 2012 – as well as other databases – to accelerate performance, improve response times, and boost efficiency.

For more information please contact www.c24.co.uk

 


Managing the flood of big data: infographic | Econsultancy

May 11, 2012

Using big data to make better decisions

By having the right data at their fingertips, marketers can make better decisions to:
•Identify high potential audiences and accurately target them
•Enable the right message at the right time via the right content targeting
•Maximize ad inventory by identifying high-value audiences across publisher properties
•Optimize ad media purchase and understand the value of channels higher up in the funnel


Follow

Get every new post delivered to your Inbox.

Join 753 other followers