Rackspace Hosting on Tuesday unveiled its new, rearchitected public cloud in an effort to deliver more flexibility for customers based upon the needs of their applications, whether those needs involve basic web hosting or massive cloud-scale databases. \n\nThe new architecture is based on what Rackspace calls Performance Cloud Servers, designed with 100 percent data center-grade RAID 10-protected solid-state disks (SSDs), Intel Xeon E5 processors, up to 120 GB of RAM and 40 Gigabits per second of highly available network throughput to the host. The goal is to offer a hosting platform that can handle workloads ranging from basic web hosting to large-scale NoSQL data stores like MongoDB and Cassandra. Performance Cloud Servers are powered by OpenStack, enabling programmatic and on-demand access. They can connect to dedicated bare metal servers as part of the Rackspace Hybrid Cloud. "Private cloud is growing as people become more comfortable with the idea of cloud," says Scott Sanchez, director of Strategy at Rackspace. "We are seeing strong growth in the private cloud. But a lot of the reason that people picked private cloud is performance. They want to be able to dictate the type of disk and the type of network. We've widened the continuum of what fits in the public cloud. "The idea, Sanchez explains, is not to convert private cloud customers to public cloud, but to offer customers the flexibility of choice. "There still is a fairly large number of organizations that are going to feel more comfortable with a private cloud," he says. "To us, they're all just building blocks. Hybrid cloud is not a product that we sell. It's an outcome of how customers want to build their clouds. We want to expand the number of customers who can look at the public cloud." The new architecture allows Rackspace to offer its public cloud in two performance "flavors," though Erik Carlin, director of the Cloud Compute Product Line at Rackspace, notes that additional flavors are in the works. "To offer performance and really run things well, we have to offer a workload-oriented view of things," Carlin says. "We developed a workload framework and a set of workloads that we wanted to work well in the cloud.""We are looking at additional classes of compute that we will deliver in the future," he adds. "We're really solving for two workloads at present: The first use case is a general-purpose node and the second use case is really solving for databases in the cloud. We have additional classes we're actively at work on now and we'll continue to iterate to deliver additional cases in the future." The first use case is served by Rackspace's new Performance 1 Flavor Class offering. It's best suited for web servers, batch processing, network appliances, small databases and other general-purpose computing workloads. It offers low-to-medium RAM, high-performance RAID 10-protected SSD storage, redundant 10 Gigabit networking, from one to eight virtual CPUs and disk I\/O up to about 35,000 4K random read IOPS and random write IOPS. The second use case is served by the Performance 2 Flavor Class offering. It's best suited for apps that demand higher RAM, more disk I\/O and more consistent performance\u2014for instance, medium to large relational databases, NoSQL data stores and distributed caches. It offers medium-to-high RAM, high-performance RAID 10-protected SSD storage, redundant 10 Gigabit networking, from four to 32 virtual CPUs and disk I\/O that scales with the number of data disks up to about 80,000 4K random read IOPS and about 70,000 4K random write IOPS. "In the old world, we were limited by the number of transactions per second SQL Server could handle," says Matthew Swanson, chief software architect for real estate lead generation specialist Commissions Inc. "With Performance Cloud, we saw upwards of 3,500 peak transactions per second. In less than 15 minutes I was able to parse through more than 6 GB of log files and put just over 1.5M rows in our SQL Server database. Performance Cloud Servers are fast\u2014very fast."Thor Olavsrud covers IT Security, Big Data, Open Source, Microsoft Tools and Servers for CIO.com. Follow Thor on Twitter @ThorOlavsrud. \nFollow everything from CIO.com on Twitter @CIOonline, Facebook, Google + and LinkedIn.