This report examines the current landscape of modern database platforms, and investigates the requirements for handling omnistructured data with a unified platform.

Key findings in this report include:

• Proliferation of data formats and platforms forces organizations to invest heavily in complex architectures so that they can process their data efficiently.
• SQL and webscale can blend, as the progress in multicore processors, parallel algorithms, and in-memory processing allows relational SQL databases to evolve and catch up to their NoSQL counterparts. The enhanced role of semistructured data should not force jettisoning of adopted query languages and technologies, since these represent a significant investment.
• Transactional and analytical workloads can be converged alongside structured and semistructured data workloads. The key lies in the multipurpose storage of data that enables this convergence. Platforms leveraging in-memory architecture can be redesigned based on data compression and indexing in-memory to deliver performance gains by orders of magnitude that enable use cases not previously possible.

Image courtesy of teekid/iStock.

The post Handling Omnistructured Data with a Unified Platform appeared first on Gigaom.

If the learning curve for reporting isn’t low, adoption will be. Reporting and analytics must be embedded inside applications and end-users must be able to use their interfaces not just for running reports but also for designing them. Highly complex reports shouldn’t take two expensive developers to build if one single client services person can do it instead.

This report is for executives, product managers and developers at independent software vendors (ISVs), SaaS vendors, solutions providers, or anyone building business applications. It will investigate how reporting needs, capabilities, and implementation requirements have changed, and present a new set of “do’s and don’ts” for successful embedded analytics.

Key findings in this report include:

• It is imperative to embed modern BI (reporting, dashboards, and visualizations) in the application users’ daily work. A generation of users that has been brought up on mobile and browser-based applications that are self-contained is unlikely to accept switching to a separate application to access BI features.
• Democratizing access to data is key, and can be achieved by enabling access to transactional databases (or read-only copies) instead of creating separate analytical databases. In doing so, the burden on IT or the data science team is minimized, and users are empowered to access data on their own.
• Democratized access to data should not equal a lack of control. There must be a mechanism for user authorization and access rights, as access to the reports and data should be restricted to only the appropriate users for any given scenario.
• Reporting functionality should have a professional look and feel that is natural for users. Reports should be easy to create, embedded in applications, and have production-level polish.
• ISVs and solution providers now face a build-versus-buy strategy for BI implementation. Arguments tend to favor “buy” because these organizations lack core BI expertise and often spend too much time and too many resources building and maintaining it in-house.

Thumbnail image courtesy of Sergey Nivens/iStock.

The post Embedded Analytics in the Self-Service BI Enterprise appeared first on Gigaom.

This report will investigate each of the three options on this spectrum, analyze the benefits and disadvantages of each, and help IT executives and database administrators considering Hadoop adoption or a change to their current processes.

Key findings in this report:

• Organizations leveraging on-premises Hadoop can control their deployments and apply optimal configuration for their workloads. This eliminates recurring infrastructure costs and any worries about data leaving the premises. However, building and maintaining this kind of in-house infrastructure requires an up-front investment as well as expertise, and using it optimally is not always possible.
• Organizations leveraging Hadoop in the cloud can achieve insight more quickly because they benefit from elasticity and do not have the burden of building and maintaining the infrastructure and developing the expertise required for supporting in-house Hadoop clusters. But costs can accumulate quickly, networking and latency usually prevent the cloud from being optimal for large workloads, and the organization is required to cede control to third parties.
• Organizations can benefit from a management layer on top of Hadoop that enables them to have the best of both approaches, but this layer must include advanced enterprise security and systems-management features.

Thumbnail image courtesy of flickr user RachScottHalls.

The post The best approach to managing Hadoop appeared first on Gigaom.

In this report, we take a look at the ingredients that make a good analytics solution: visualization, self-service tools, and guided machine learning. We identify the strengths and weaknesses that each of these ingredients brings when used individually, and we show why combining these strengths into a new approach is essential for an efficacious, comprehensive solution.

Key findings of this report include:

• Self-service tools using visualization may have shortened the delivery cycle for data analysis and in certain cases allowed business users to take matters into their own hands. However, without proper backing, self-service analytics is a double-edged sword, as it provides a false sense of confidence that can lead to statistically unsupported conclusions.
• Some may see data scientists as the new rock stars with organizations putting a burden on their shoulders to deliver everything they hope their data analysis can be. But the all-encompassing notion of a data scientist is inaccurate. Real data scientists are in short supply, and placing such strong dependency on one person or a small team is a risky proposition for any business.
• Machine learning is a process that has been hard and expensive to configure and execute. Even experts have a hard time interpreting generated results. Business users and experts need more-comprehensible self-service machine learning.

Thumbnail image courtesy of: iStock/Thinkstock.

The post Approachable analytics: data intelligence for everyone appeared first on Gigaom.

Key findings include:

• Traditional application architecture is a patchwork that is unable to cope with the challenges of today’s business needs.
• An agile and holistic approach is needed, and applications cannot possibly support this by operating in silos; integration is the key.
• If we move applications to the cloud without a change of architectural paradigm, we will at best get modest benefits and at worst broken applications. We need a new set of guidelines to build applications suitable for today’s business.

The post Second-generation cloud architecture: breaking the application silo appeared first on Gigaom.

However, as key Hadoop vendors are pointing out, it normally takes about a decade for a new market to consolidate and reach maturity status, and Hadoop is no exception. It has been around since 2005, and one of the areas of this rapidly growing and maturing ecosystem that remains lackluster is that of security.

The problem stems from Hadoop’s origins: It is a platform designed and developed primarily for data-specialist use-case scenarios. Hadoop hits its mark best when used by small, cohesive teams of experts working in isolated environments on data sets explicitly assigned to them and under their control.

But now that Hadoop’s application is transitioning increasingly to that of serving as a backbone for enterprise data management those implicit assumptions are no longer valid. In fact, this very realization has led the evolution of Hadoop from its initial design based on the MapReduce batch file-processing paradigm to its more recent incarnation, broadening its scope to accommodate other data processing approaches.

Thumbnail image courtesy of vitanovski/Thinkstock.

The post Hadoop security: Solutions emerge appeared first on Gigaom.

This report explores the key features that influence the evolution of agile BI and takes a look at the BI landscape under this light. At first glance, polarization seems to exist between traditional BI vendors, who are focused on extract, transform, and load (ETL) and reporting, and the newcomers, who are focused on data exploration and visualization, but a closer look reveals that, in fact, they converge as adoption of useful features is taking place across the spectrum.

This report will illustrate for both the traditional BI vendors and the newcomers that:

• As the market is expanding, features such as cloud support and embedded domain-specific knowledge in BI solutions are key. Initially, the benefits will be more obvious to those smaller players who do not have the resources for in-house infrastructure and extended internal projects and who are driven more by needing immediate results. Over the long run, however, these features can benefit all types of organizations.

• Ubiquity and mobility are key features of data today; therefore, the ability to support a multitude of data sources with as little effort as possible – integrating them and accessing analysis results via a multitude of channels – is important in order to keep up.

• We are shifting from static reports to interactive visualization. The focus is also shifting from having an overview of metrics to being able to discover what are the causes and effects of the phenomena the metrics express.

Thumbnail image courtesy of SirkulT/Thinkstock.

The post Agile business intelligence: reshaping the landscape appeared first on Gigaom.