Boosting App Performance in the Cloud

When enterprises use the cloud as a storage platform, the main challenge is to ensure reliability and availability at a level users have grown accustomed to.

But now that more and more cloud deployments are being used to support application-layer infrastructure, the challenges increase by orders of magnitude. Not only does the cloud need to be up and running at all times, but it has to perform at a level that is not always easy to maintain, or to even monitor.

That’s why we’re seeing a wave of cloud-ready application performance management (APM) solutions in the channel. It seems that the more distributed data architectures become, the greater the need for high visibility and greater coordination among disparate architectures. And the market seems ripe for a new breed of startups that are looking at the data environment in entirely new ways.

One of them is Ipanema Technologies, which recently released v7.1 of its Autonomic Network System (ANS) platform. The system kicks the notion of APM scalability to new levels, offering up to 20,000 new connections per second for improved high-end device performance. That pushes throughput up to 2 Gbps and enables more than 300 applications to be classified upon start-up. That, along with intelligent device clustering and advanced deep packet inspection enables app performance of up to 10 Gbps. It also provides cloud application monitoring (CAM) reporting and virtual machine deployment suitable for all-cloud or hybrid distribution.

There are also a number of APM platforms tailored to specific cloud environments. SL Corp., for instance, has added a new Amazon CloudWatch adapter to its RTView Enterprise Monitor, which gives the platform a number of new tools aimed at tracking application performance on Amazon Web Services. The package includes the RTView Historian that stores numerous metrics in a SQL database to conduct trend analyses and define alert thresholds to head off issues before they impact performance. As well, the system incorporates app performance data from sources like the application server or message bus into a single-view platform for real-time management of utilization, latency and other parameters.

A number of firms are also providing agentless APM cloud services, which enable broader management of applications on virtual platforms without having to distribute multiple software modules (the agents) throughout on-premises infrastructure. ExtraHop, for example, has released a new agentless platform for Hyper-V environments that enables visibility across multiple application, network and storage environments. The system uses the Hyper-V Virtual Switch function to track performance across public and private infrastructure in the Windows Azure cloud, even as the number of virtual servers extends into the multi-thousands.

As well, many of the traditional WAN optimization developers are turning their attention to APM as application traffic grabs ever-larger portions of the links to various cloud services. Blue Coat recently updated its PacketShaper appliance with new tools to increase visibility into web applications and extend greater control over policies and deployment of enterprise services. One of the goals is to combat the rise of “shadow infrastructure” as employees take it upon themselves to spin up not only unauthorized cloud infrastructure but new application sets, which could have a profound impact on overall data performance. The new system provides a more detailed view of web, native and mobile applications as well as user activity across multiple devices, allowing for the establishment of more accurate service-level and other usage policies.

As virtualization and the cloud add layers of abstraction between data, applications and underlying hardware, management will continue to shift away from server, storage and networking performance and more toward the application/service layer. For the enterprise, at least, that means a shift of focus away from identifying and repairing bottlenecks to rerouting traffic around the weak spots in physical infrastructure.

Successful IT managers, then, will be judged by their ability to improve operations, not just spot and fix problems.