Driving the New Wave of High Performance E-Commerce Applications

Many mid-sized e-commerce companies can win or lose in today’s hyper-competitive market based on how well they use their data—both in day-to-day operations and in analytics. In this paper, learn how a powerful NewSQL database, ClustrixDB, is key to bringing the performance of distributed computing to e-commerce applications, giving companies a competitive advantage right when it’s needed most.

How NewSQL Delivers Speed, Scale, and Simplicity

Introduction

Relational database management systems (RDBMS) such as MySQL have historically served as the central repository for critical information in mid-sized e-commerce organizations. However, many of today’s data-intensive e-commerce applications have exposed the uncomfortable truth that traditional RDBMS architectures have reached their limits. They are no longer able to keep pace with the demands of the modern enterprise.

These e-commerce applications all have a number of common traits:

  • Large scale systems, with huge – and growing – product catalogs and customer data
  • An underlying relational data model
  • Transactional integrity is important, and consistency matters
  • High concurrency requirements

NoSQL databases have made a big splash in the market recently. And while their scale-out architectures address the issue of large and fast growing data sets, they lack the transactional integrity and ACID guarantees that business critical e-commerce applications expect. Fortunately, NewSQL databases have emerged bringing the power of distributed computing to RDBMS data and related applications.

In this white paper, we introduce the ClustrixDB distributed, fault-tolerant e-commerce database, which enables organizations to protect their investments in relational data and associated application logic while delivering superior responsiveness, scale, and simplicity. We begin by explaining why legacy relational database architectures are insufficient for today’s needs. Next, we summarize the organizational, operational, and financial implications of this shortfall and describe why a fresh approach to the relational database is essential. Finally, we introduce ClustrixDB and highlight its capabilities and benefits.

The intended audience includes IT and business executives, application architects and developers, and database architects and administrators with responsibility for developing and maintaining e-commerce applications.

Big data breaks legacy databases

Traditional Database Architectures Have Reached their Limit

Today, every e-commerce company confronts a highly complex, rapidly changing IT landscape. First, there are new channels to support, including social and mobile. Next, new and existing e-commerce applications alike are generating, capturing, and managing much more information than ever before. This includes more classes and categories of data. Furthermore, driven by sophisticated business intelligence and analytic tools, information is being retained and utilized for longer periods of me. User expectations for up me, performance, functionality, and accessibility are also much higher now. Finally, it is common for businesses to compete – either explicitly or implicitly – on the quality and competence of their websites.

Today’s e-commerce websites must find a way to cope with two contradictory realities:

  1. Big Data overwhelms legacy RDBMS technology. Data volumes, transactional loads, and concurrency requirements have outstripped the 40-year-old architecture behind traditional RDBMS.
  2. Yet the vast majority of data continues to be stored in RDBMS. There are many reasons for this fact, including:
    • The value and meaning of all this data is based on its relations, such as among customers, orders, inventory, and so on.
    • Transactional integrity via ACID (Atomicity, Consistency, Isolation, Durability) compliance is crucial for core e-commerce applications.
    • There is an enormous installed base of RDBMS developer expertise.

Core e-commerce transactions depend on operations that are driven by inter-related data.

In response to all these burdens – and in an effort to reconcile the paradox – e-commerce application developers and solution providers have tried a variety of approaches to coaxing more from their information infrastructure. These measures include:

  • Scaling up hardware. This approach – which entails stuffing faster processors, more memory, and extra disk storage into a single, powerful server – can be extremely expensive, and the law of diminishing returns sets in quickly: at best, these measures only defer the reckoning.
  • Sharding. This technique of data distribution is highly complex to design and administer. It requires significant internal expertise and demands specialized application logic (bolt on so ware that sits in front of the database is popular, but adds complexity).
  • NoSQL. While these new classes of database are useful for certain specialized classes of information and use cases, they are not suitable for the vast majority of relational data. For example, NoSQL is not meant for e-commerce transactions. It also requires extensive application re-writes.

None of the above tactics are core competencies for most organizations. All consume extra time and money, and all require specialized proficiencies. The net result is diminished innovation and delayed application delivery, which directly leads to a weakened competitive stance. Worst of all, after wasting abundant me and money trying to build and maintain risky and complex database plumbing, the business isn’t likely to get the solution it really needs.

Since merchants are clamoring for standardized, vendor-supplied solutions that scale, the NewSQL revolution is starting for their most precious corporate IT asset of them all: the primary, relational data store that is the backbone of the vast majority of e-commerce applications. As we show in figure 3, modern data management architectures will include scale-out on commodity hardware in the cloud, and other technologies that must all work together.

Database Management System Image

Introducing ClustrixDB

The ‘NewSQL’ term refers to a confluence of factors that now make it possible to preserve investments in relational data and application logic, yet still deliver the performance, reliability, and scalability demanded by today’s e-commerce applications. These dynamics include:

  • The LAMP (Linux, Apache, MySQL, PHP/Perl) stack
  • Innovation in distributed, parallel relational database so ware
  • Cloud computing and Database as a Service (DBaaS)

Clustrix is the leading supplier of NewSQL solutions. Clustrix was founded to bring the benefits of NewSQL to transactional, e-commerce applications by providing a database solution that:

  • Delivers linear scale to accommodate unlimited numbers of users, transactions, and data by leveraging the power of massively parallel processing.
  • Protects information with full ACID capabilities, thereby ensuring transactional integrity and consistency.
  • Supplies automatic fault tolerance, freeing developers and administrators to concentrate on fulfilling their primary responsibilities.
  • Deploys quickly and easily, with both installation and expansion requiring only minutes to complete.

All without requiring complex, challenging DBA-level workarounds such as sharding. Available as a so ware download for popular Linux distros as well as in public, private, and hybrid-cloud configurations, Clustrix’ technology strategy provides three principal technical advantages:

  1. Data distribution. Available as either a cloud or on-premise so ware solution, the ClustrixDB pla orm implements a shared-nothing architecture and is constructed using industry-standard building blocks. ClustrixDB automatically distributes information across nodes to provide the best possible performance, without the need for any database administrator intervention or specialized application logic. Flexible, on-line schematichanges allow agile so ware delivery.
  2. Massively parallel query processing. The distributed relational database architecture of ClustrixDB provides linear transactional performance. Nodes can be added as needed, and there is no upper limit on the number of nodes that can be deployed. Adding nodes doesn’t require any application or administrative alterations. ClustrixDB provides a single query interface that lets users submit queries across all nodes without needing to know where data resides. Instead, queries are brought to the data and run in parallel.
  3. Easy administration. As a MySQL-compatible solution, ClustrixDB automatically translates MySQL queries to searches that are optimized for the ClustrixDB distributed architecture. This lets developers build applications using the most widely used SQL language, yet still realize the benefits of the ClustrixDB approach. Automatic fault tolerance ensures continuous application availability with built-in redundancy and failover at the data level; online schema changes enable application innovation without needing to take the database offline. Clustrix support remotely monitors database performance and proactively engages customers with recommendations for performance and scalability.

Two key innovations in the Clustrix distributed architecture.

Conclusion

Although coming up with specific, accurate technology predictions is a formidable task, there are a few certainties: e-commerce applications will continue to grow in size, scale, and complexity, and data volumes will keep getting larger. The vast majority of this information will still need to be stored in relational databases, yet the customary relational database model cannot deliver the performance and scalability mandated by all of this data.

Far too many e-commerce companies are squandering valuable me, talent, and money trying to create and support ad-hoc, one-o solutions to cope with these demands. NewSQL databases represent an ideal approach to carry the existing relational model forward and properly support the next generation of applications. Clustrix provides a market-proven NewSQL solution that delivers speed, scalability, and simplicity for enterprises that want to protect their investments in relational databases and applications, yet still keep up with today’s requirements.