Introducing Cisco UCS Common Platform Architecture (CPA) for Big Data

October 31, 2012 - 0 Comments

Updated: 10/01/2013

You may have heard that the digital universe is in petabytes, global IP traffic is in 100s of exabytes. These are mind bogglingly large metrics. Big data analytics can play a crucial role in making datasets in this space usable – by improving operational efficiency to customer experience to prediction accuracy. While Cisco is the global leader in networking – Did you know that 85% of estimated 500 exabyte global IP traffic in 2012 will pass through Cisco devices ? – the company also builds an innovative family of unified computing products. This enables the company to provide a complete infrastructure solution including compute, storage, connectivity and unified management for big data applications that reduce complexity, improves agility, and radically improves cost of ownership.

To meet a variety of big data platform demands (Hadoop, NoSQL Databases, Massively Parallel Processing Databases etc), Cisco offers a comprehensive solution stack: the Cisco UCS Common Platform Architecture (CPA) for Big Data includes compute, storage, connectivity and unified management. Unique to this architecture is the seamless data integration and management integration capabilities with enterprise application ecosystem including Oracle RDBMS/RAC, Microsoft SQL Server, SAP and others. See Figure 1.

Figure 1:

The Cisco UCS CPA for Big Data is built using the following components:

  • Cisco UCS 6200 Series Fabric Interconnects provides high speed, low latency connectivity for servers and centralized management for all connected devices with UCS Manager. Deployed in redundant pairs offers the full redundancy, performance (active-active), and exceptional scalability for large number of nodes typical in big data clusters. UCS Manger enables rapid and consistent server integration using service profile, ongoing system maintenance activities such as firmware update operations across the entire cluster as a single operation, advanced monitoring, and option to raise alarms and send notifications about the health of the entire cluster.
  • Cisco UCS 2200 Series Fabric Extenders, act as remote line cards for Fabric Interconnects providing a highly scalable and extremely cost-effective connectivity for large number of nodes.
  • Cisco UCS C240 M3 Rack-Mount Servers, 2-RU server designed for wide range of compute, IO and storage capacity demands. Powered by two Intel Xeon E5-2600 series processors and support up to 768 GB of main memory (typically 128GB or 256GB for big data applications) and up to 24 SFF disk drives in the performance optimized option or 12 LFF disk drives in the capacity optimized option. Also features Cisco UCS VNIC optimized for high bandwidth and low latency cluster connectivity with support for up to 256 virtual devices.

The Cisco UCS CPA for Big Data meets diverse big data requirements and support leading software distributions from our ecosystem partners including Cloudera, Intel, Greenplum, MapR, MarkLogic, Oracle, ParAccel  and others. Cisco UCS is the exclusive hardware reference platform for Intel Hadoop,  MapR and ParAccel, and exclusive partner platform for Oracle NoSQL Database. Deep technology integration and engineering collaboration with these partners enable us to offer complete solution allowing customers to shift their focus from custom building the solutions and constant maintenance to strategic business initiatives.

Current version of the Cisco UCS CPA for Big Data offers two options: a high-performance and a high-capacity – depending on the compute and storage requirements. The high-performance option offers balance of compute power with IO bandwidth optimized for price/performance is built using C240 M3 rack servers powered by two Intel Xeon E5-2665 processors (16 cores) with 256 GB of memory and 24 1TB SFF disk drives. The high-capacity option optimized for low cost per terabyte is built using C240 M3 rack servers powered by two Intel Xeon E5-2640 processors (12 cores) with 128GB of memory and 12 3TB LFF disk drives.

The solutions are offered in single rack and multiple rack scale. The single rack configuration consists of two Cisco UCS 6248UP 48-Port Fabric Interconnects (supports  up to five racks) or Cisco UCS 6296UP 96-port Fabric Interconnects (supports up to 10 racks, 160 servers) along with two Cisco Nexus 2232PP 10GigE Fabric Extenders and 16 Cisco UCS C240 M3 Rack-Mount Servers (high-performance or high-capacity models). Each server in the configuration connects to the unified fabric through two active-active 10-GigE links using a Cisco UCS VNIC. Multi-rack configurations include two Cisco Nexus 2232PP fabric extenders and 16 Cisco UCS C240 M3 Rack-Mount Servers for every additional rack. The high performance rack can support up to 256 cores and 32 Gbytes/sec (SATA) or 48 Gbytes/sec (SAS) IO bandwidth, high capacity rack can support up to 576 TB raw storage (that’s 5.7 PB per UCS domain).

It’s exciting to see that within only a few months of making the Cisco UCS CPA for Big Data solutions available they are already being deployed in a range of industries including finance, retail, service provider, content management and government.

The high performance rack and high capacity rack are available through the Cisco SmartPlay program in single SKU bundles as shown in figure 2.

Figure 2:
Cisco SmartPlay Solution Bundles

Scaling beyond 10 racks (160 servers) can be implemented by interconnecting multiple UCS domains using Nexus 6000/7000 Series switches, scalable to thousands of servers and to hundreds of petabytes storage, and managed from a single pane using UCS Central in a datacenter or distributed globally as shown in figure 3.

Figure 3:
Big Data UCS Central

White Papers

[1] Cisco UCS with the Intel Distribution for Apache Hadoop
[2] Cisco UCS with MapR: Delivering Advanced Performance for Hadoop Workloads
[3] Cisco and Greenplum Partner to Deliver High-Performance Hadoop Solution
[4] Cisco UCS Ecosystem for Oracle: Extend Support to Big Data and Oracle NoSQL Database
[5] Cloudera Enterprise with Cisco Unified Computing System
[6] Cisco Big Data Common Platform Architecture (CPA) : Announcing SmartPlay Solution Bundles
[7] Cisco UCS with ParAccel Analytic Platform Solution: Deliver Powerful Analytics to Transform Business
[8] Extreme Performance (Million OPS) to Oracle NoSQL Database Big Data Applications
[9] Pentaho High-Performance Big Data Reference Configuration Using Cisco UCS
[10 ] Cisco UCS with StackIQ Solution: Deliver Big Infrastructure for Big Data

Cisco Big Data Design Zone

In an effort to keep conversations fresh, Cisco Blogs closes comments after 60 days. Please visit the Cisco Blogs hub page for the latest content.