ATCA DPI Framework
DPI Framework for ATCA Platforms
To deliver optimal performance and efficiency, network equipment providers are making increased use of Deep Packet Inspection (DPI) technology to identify applications and network usage. DPI can deliver real-time information about the network, thereby enabling a variety of new applications including policy enforcement, Internet offload gateways, video optimization, edge caching, RAN network monitoring and security amongst others.
As DPI transitions from specialized to mainstream “must have” solutions, equipment manufacturers are looking for additional support to develop and deliver these advanced capabilities – and often these technologies are beyond their traditional core expertise.
Fortunately, the DPI software framework brings together a range of best-in-class components as part of a robust framework to deliver DPI capabilities for these new generations of applications. For many applications, the DPI software framework will provide a stateful session-based load balancer with capability to filter, forward and manipulate bearer plane traffic at line rates up to 80Gbps per blade...equating to capacities of 200Gbps+ per system.
Stateful load balancers, filtering and forwarding of traffic are a fundamental part of building discrete packets into coherent flows (sessions) which can then be analyzed to extract information such as application type, user ID and other specifics including metadata. Once analyzed, policies can be easily created and applied to all subsequent packets that are part of this session.
For example, traffic matching pre-defined criteria can be separated and re- directed for offloading or video compression. Alternatively, the DPI software framework instances can be cascaded to sort and filter traffic prior to applying policies or forwarding for further analysis and processing. Use cases include lawful intercept solutions and standalone PCEFs in the mobile network – or PCEFs as part of mobile core network gateways.
FLOWENGINE CORE FRAMEWORK
- State table with 1M+ flow entries
- Pre-processing functions including de-tunneling & de-fragmentation
- 5-tuple-based key extraction for look-up
PRE-PROCESSING AND POLICY EXTENSIONS
- 3G & LTE specific protocols and extensions (GTP-U)
- Time-stamping and packet re-ordering
- Rules engine support for VLAN tags, forwarding, TCP intercept/re-direct
CLASSIFICATION ENGINE
- Basic ID support based on L4 parameters, e.g., GTP APN, TEID
- DPI-based detection of application and user
- Extended DPI analysis to extract application metadata, e.g., User ID
HARDWARE PLATFORM
- ATCA 40G and 10G platforms
- Scales from 20G to 200G+
- Core FlowEngine functions ported to ATCA-PP81
- Optional Intel x86-based DPI content extraction and processing
HARDWARE PLATFORM
- 40G ATCA hub switch with 10G, 40G and 100G uplinks
- ATCA-PP81 packet processor blade using dual XLP832 processors
RULES ENGINE
- Packet matching known session in flow table can have standard rule applied
- Custom rules can be created and added as required including wireless control plane extensions
- Add VLAN tag – Forward packet
- Drop packet
- TCP session re-direct – Qos / Rate limiting (OEM)
CLASSIFIER MODULE
- Packets not matching flows already in the flow table are routed to the classification engine
- Sequences of packets can be matched to known application signatures or specific header parameters (L4) to uniquely identify session. Identification criteria can include:
- HTTP traffic & content type "video": applicable to video optimization gateways
- GTP-based header field(s): Internet offload
- Applications (up to 300+ individual signatures): based on Qosmos IxEngine
FLOW TABLE & KEY GENERATION
- 1M+ flow table entries; scalable memory for higher numbers of flows
- Key generation based on embedded user packet
- 5-tuple-based hash including source & destination IP address
PACKET PRE-PROCESSING FUNCTION
- Packet parsing engine
- Including IP-IP, L2TP, GTP, MPLS, etc.
- Defragmentation
- Time stamping
PACKAGING
- ATCA platform scalable from 2 to 14 slots
- FlowEngine ported and designed to run on XLP-based PP81
- Portable to ATCA-XE80 blades with Intel CPUs
- Compatible with load balancing software on 40G hub switch - Deliverables – FlowEngine framework delivered as source code
- Includes pre-defined APIs for standard functions
- Extensible by user or Trillium Professional Services
- Third party classification and pre-processor modules delivered as binaries - Option to license full development tools directly from third parties
- Option to license as application-specific pre-configured binary if required
For more information
(800) 950-0044
For support questions call 1-866-385-6167 or visit support downloads for this product