SlideShare a Scribd company logo
An Introduction to Open Fabric Interfaces
Abstract: OFI are high-performance software interfaces that provide low-latency access to fabric
hardware.
Architectural Overview
Open Fabric Interfaces, or OFI, are a set of application programming interfaces (APIs) that export
communication services to applications. OFI is specifically designed to meet the performance and
scalability requirements of high-performance computing (HPC) applications, such as MPI, SHMEM, PGAS,
DBMS, and enterprise applications, running in a tightly coupled network environment. OFI is agnostic to
the underlying networking protocols, as well as the implementation of the networking devices.
OFI is instantiated in the libfabric library. Libfabric is supported on commonly available Linux based
distributions.
OFI is architected to support process direct I/O. Process direct I/O, historically referred to as RDMA,
allows an application to access network resources without operating system interventions. Data
transfers can occur between networking hardware and application memory with minimal software
overhead. Although OFI supports process direct I/O, it does not mandate any implementation or require
operating system bypass.
The following diagram highlights the general architecture of the interfaces exposed by OFI. For
reference, the diagram shows OFI in reference to a NIC. This is provided as an example only of how
process direct I/O may be supported.
OFI Architecture
OFI is divided into two separate components. The main component is the OFI framework, which defines
the interfaces that applications use. The OFI frameworks provides some generic services; however, the
bulk of the OFI implementation resides in the providers. Providers plug into the framework and supply
access to fabric hardware and services. Providers are often associated with a specific hardware device
or NIC. Because of the structure of the OFI framework, applications access the provider implementation
directly for most operations, in order to ensure the lowest possible software latencies.
As captured in the architecture diagram, OFI can be grouped into four main collections of interface sets.
Control Services: These are used by applications to discover information about the types of
communication services are available in the system. For example, discovery will indicate what fabrics
are reachable from the local node, and what sort of communication each provides.
Communication Services: These interfaces are used to setup communication between nodes. It
includes calls to establish connections (connection management), as well as functionality used to
address connectionless endpoints (address vectors).
Completion Services: OFI exports asynchronous interfaces. Completion services are used to report the
results of submitted operations. Completions may be reported using event queues, which provide
details about the operation that completed. Or, completions may be reported using lower-impact
counters that simply return the number of operations that have completed.
Data Transfer Services: These are sets of interfaces designed around different communication
paradigms. Although shown outside the data transfer services, triggered operations are strongly related
to the data transfer operations.
There are four basic data transfer interface sets. Message queues expose the ability to send and receive
data with message boundaries being maintained. Message queues act as FIFOs, with sent messages
matched with receive buffers in the order that messages are received. Tag matching is similar to
message queues in that it maintains message boundaries. Tag matching differs from message queues in
that received messages are directed into buffers based on small steering tags that are carried in the sent
message.
RMA stands for remote memory access. RMA transfers allow an application to write data directly into a
specific memory location in a target process, or to read memory from a specific address at the target
process and return the data into a local buffer. Atomic operations are similar to RMA transfers, in that
they allow direct access to the memory on the target process. Atomic operations allow for manipulation
of the memory, such as incrementing the value found in memory. Because RMA and atomic operations
provide direct access to a process’s memory buffers, additional security synchronization is needed.
Object Model
Interfaces exposed by OFI are associated with different objects. The following diagram shows a high-
level view of the parent-child relationships.
OFI Parent-Child Object Relationships
Fabric: A fabric represents a collection of hardware and software resources that access a single physical
or virtual network. For example, a fabric may be a single network subnet. All network ports on a system
that can communicate with each other through the fabric belong to the same fabric domain. A fabric
shares network addresses and can span multiple providers.
Domain: A domain represents a logical connection into a fabric. For example, a domain may map to a
physical or virtual NIC. A domain defines the boundary within which fabric resources may be associated.
Each domain belongs to a single fabric.
Passive Endpoint: Passive endpoints are used by connection-oriented protocols to listen for incoming
connection requests. Passive endpoints often map to software constructs and may span multiple
domains.
Event Queues: EQs are used to collect and report the completion of asynchronous operations and
events. Event queues handle events that are not directly associated with data transfer operations,
referred to as control events. For example, connection requests and asynchronous errors that are not
associated with a specific data transfer are reported using event queues.
Wait Sets: Although OFI allows the use of native operating system constructs for applications to use
when waiting for events, it defines an optimized method for applications to use when waiting on events
across multiple event queues, completion queues, and counters. Wait sets allow a single underlying
wait object to be signaled whenever a specified condition occurs on an associated event queue,
completion queue, or counter.
Active Endpoint: Active endpoints are data transfer communication portals. Active endpoints are used
to perform data transfers, and are conceptually similar to a socket. Active endpoints are often
associated with a single hardware NIC.
Completion Queue: Completion queues are high-performance queues used to report the completion of
data transfer operations. Unlike fabric event queues, completion queues are often associated with a
single hardware NIC.
Completion Counter: Completion counters are used to report the number of completed data transfer
operations. Completion counters are considered lighter weight than completion queues, in that a
completion simply increments a counter, rather than placing an entry into a queue.
Poll Set: OFI allows providers to use an application’s thread to process asynchronous requests. This can
provide performance advantages for providers that use software to progress the state of a data transfer.
Poll sets allow an application to group together multiple objects, such that progress can be driven across
all associated data transfers. In general, poll sets are used to simplify applications where manual
progress model is employed.
Memory Region: Memory regions describe application’s local memory buffers. In order for fabric
resources to access application memory, the application must first grant permission to the fabric
provider by constructing a memory region. Memory regions are required for specific types of data
transfer operations, such as RMA and atomic operations.
Address Vectors: Address vectors are used by connectionless endpoints. They map higher level
addresses, such as IP addresses, which may be more natural for an application to use, into fabric specific
addresses. The use of address vectors allows providers to reduce the amount of memory required to
maintain large address look-up tables, and eliminate expensive address resolution and look-up methods
during data transfer operations.
Connection-Oriented Communication
OFI supports both connected and connectionless communication. The following diagram highlights the
general usage behind connection-oriented communication.
Connected Endpoints
Connections require the use of both passive and active endpoints. In order to establish a connection, an
application must first create a passive endpoint and associate it with an event queue. The event queue
will be used to report the connection management events. The application then calls listen on the
passive endpoint. A single passive endpoint can be used to form multiple connections.
The connecting peer allocates an active endpoint, which is also associated with an event queue.
Connect is called on the active endpoint, which results in sending a connection request (CONNREQ)
message to the passive endpoint. The CONNREQ event is inserted into the passive endpoint’s event
queue, where the listening application can process it.
Upon processing the CONNREQ, the listening application will allocate an active endpoint to use with the
connection. The active endpoint is bound with an event queue. Although the diagram shows the use of
a separate event queue, the active endpoint may use the same event queue as used by the passive
endpoint. Accept is called on the active endpoint to finish forming the connection. It should be noted
that the OFI accept call is different than the accept call used by sockets. The differences result from OFI
supporting process direct I/O.
OFI does not define the connection establishment protocol, but does support a traditional three-way
handshake used by many technologies. After calling accept, a response is sent to the connecting active
endpoint. That response generates a CONNECTED event on the remote event queue. If a three-way
handshake is used, the remote endpoint will generate an acknowledgement message that will generate
a CONNECTED event for the accepting endpoint. Regardless of the connection protocol, both the active
and passive sides of the connection will receive a CONNECTED event that signals that the connection has
been established.
Connectionless Communications
Connectionless communication allows data transfers between active endpoints without going through a
connection setup process. The diagram below shows the basic components needed to setup
connectionless communication.
Connectionless Endpoints
OFI requires the addresses of peer endpoints be inserted into a local addressing table, or address vector,
before data transfers can be initiated against the remote endpoint. Address vectors abstract fabric
specific addressing requirements and avoid long queuing delays on data transfers when address
resolution is needed. For example, IP addresses may need to be resolved into Ethernet MAC addresses.
Address vectors allow this resolution to occur during application initialization time. OFI does not define
how an address vector be implemented, only its conceptual model.
Address vectors may be associated with an event queue. After an address is inserted into an address
vector and the fabric specific details have been resolved, a completion event is generated on the event
queue. Data transfer operations against that address are then permissible on active endpoints that are
associated with the address vector. Connectionless endpoints must be associated with an address
vector.
Endpoints
Endpoints represent communication portals, and all data transfer operations are initiated on endpoints.
OFI defines the conceptual model for how endpoints are exposed to applications, as demonstrated in
the diagrams below.
Basic Endpoint Architecture
Endpoints are usually associated with a transmit context and a receive context. Transmit and receive
contexts are often implemented using hardware queues that are mapped directly into the process’s
address space, though OFI does not require this implementation. Although not shown, an endpoint may
be configured only to transmit or receive data. Data transfer requests are converted by the underlying
provider into commands that are inserted into transmit and/or receive contexts.
Endpoints are also associated with completion queues. Completion queues are used to report the
completion of asynchronous data transfer operations. An endpoint may direct completed transmit and
receive operations to separate completion queues, or the same queue (not shown).
A more advanced usage model of endpoints that allows for resource sharing is shown below.
Shared Transmit and Receive Contexts
Because transmit and receive contexts may be associated with limited hardware resources, OFI defines
mechanisms for sharing contexts among multiple endpoints. The diagram above shows two endpoints
each sharing transmit and receive contexts. However, endpoints may share only the transmit context or
only the receive context or neither. Shared contexts allow an application or resource manager to
prioritize where resources are allocated and how shared hardware resources should be used.
Completions are still associated with the endpoints, with each endpoint being associated with their own
completion queue(s).
The final endpoint model is known as a scalable endpoint. Scalable endpoints allow a single endpoint to
take advantage of multiple underlying hardware resources.
Scalable Endpoints
Scalable endpoints have multiple transmit and/or receive contexts. Applications can direct data
transfers to use a specific context, or the provider can select which context to use. Each context may be
associated with its own completion queue. Scalable contexts allow applications to separate resources to
avoid thread synchronization or data ordering restrictions.

More Related Content

What's hot

NETCONF Call Home
NETCONF Call Home NETCONF Call Home
NETCONF Call Home
ADVA
 
eBPF Workshop
eBPF WorkshopeBPF Workshop
eBPF Workshop
Michael Kehoe
 
BPF & Cilium - Turning Linux into a Microservices-aware Operating System
BPF  & Cilium - Turning Linux into a Microservices-aware Operating SystemBPF  & Cilium - Turning Linux into a Microservices-aware Operating System
BPF & Cilium - Turning Linux into a Microservices-aware Operating System
Thomas Graf
 
Cisco CCNA- DHCP Server
Cisco CCNA-  DHCP ServerCisco CCNA-  DHCP Server
Cisco CCNA- DHCP Server
Hamed Moghaddam
 
Cilium - API-aware Networking and Security for Containers based on BPF
Cilium - API-aware Networking and Security for Containers based on BPFCilium - API-aware Networking and Security for Containers based on BPF
Cilium - API-aware Networking and Security for Containers based on BPF
Thomas Graf
 
OpenFlow tutorial
OpenFlow tutorialOpenFlow tutorial
OpenFlow tutorial
openflow
 
Vxlan deep dive session rev0.5 final
Vxlan deep dive session rev0.5   finalVxlan deep dive session rev0.5   final
Vxlan deep dive session rev0.5 final
KwonSun Bae
 
Linux Native, HTTP Aware Network Security
Linux Native, HTTP Aware Network SecurityLinux Native, HTTP Aware Network Security
Linux Native, HTTP Aware Network Security
Thomas Graf
 
eBPF - Observability In Deep
eBPF - Observability In DeepeBPF - Observability In Deep
eBPF - Observability In Deep
Mydbops
 
Aci presentation
Aci presentationAci presentation
Aci presentation
Joe Ryan
 
eBPF Basics
eBPF BasicseBPF Basics
eBPF Basics
Michael Kehoe
 
A comparison of segment routing data-plane encodings
A comparison of segment routing data-plane encodingsA comparison of segment routing data-plane encodings
A comparison of segment routing data-plane encodings
Gunter Van de Velde
 
ACI Hands-on Lab
ACI Hands-on LabACI Hands-on Lab
ACI Hands-on Lab
Cisco Canada
 
Open ebs 101
Open ebs 101Open ebs 101
Open ebs 101
LibbySchulze
 
Enable DPDK and SR-IOV for containerized virtual network functions with zun
Enable DPDK and SR-IOV for containerized virtual network functions with zunEnable DPDK and SR-IOV for containerized virtual network functions with zun
Enable DPDK and SR-IOV for containerized virtual network functions with zun
heut2008
 
cilium-public.pdf
cilium-public.pdfcilium-public.pdf
cilium-public.pdf
Sanjeev Rampal
 
IPMI is dead, Long live Redfish
IPMI is dead, Long live RedfishIPMI is dead, Long live Redfish
IPMI is dead, Long live Redfish
Bruno Cornec
 
Kubernetes in Hybrid Environments with Submariner
Kubernetes in Hybrid Environments with SubmarinerKubernetes in Hybrid Environments with Submariner
Kubernetes in Hybrid Environments with Submariner
Kublr
 
The TCP/IP Stack in the Linux Kernel
The TCP/IP Stack in the Linux KernelThe TCP/IP Stack in the Linux Kernel
The TCP/IP Stack in the Linux Kernel
Divye Kapoor
 
Vpc notes
Vpc notesVpc notes
Vpc notes
Krunal Shah
 

What's hot (20)

NETCONF Call Home
NETCONF Call Home NETCONF Call Home
NETCONF Call Home
 
eBPF Workshop
eBPF WorkshopeBPF Workshop
eBPF Workshop
 
BPF & Cilium - Turning Linux into a Microservices-aware Operating System
BPF  & Cilium - Turning Linux into a Microservices-aware Operating SystemBPF  & Cilium - Turning Linux into a Microservices-aware Operating System
BPF & Cilium - Turning Linux into a Microservices-aware Operating System
 
Cisco CCNA- DHCP Server
Cisco CCNA-  DHCP ServerCisco CCNA-  DHCP Server
Cisco CCNA- DHCP Server
 
Cilium - API-aware Networking and Security for Containers based on BPF
Cilium - API-aware Networking and Security for Containers based on BPFCilium - API-aware Networking and Security for Containers based on BPF
Cilium - API-aware Networking and Security for Containers based on BPF
 
OpenFlow tutorial
OpenFlow tutorialOpenFlow tutorial
OpenFlow tutorial
 
Vxlan deep dive session rev0.5 final
Vxlan deep dive session rev0.5   finalVxlan deep dive session rev0.5   final
Vxlan deep dive session rev0.5 final
 
Linux Native, HTTP Aware Network Security
Linux Native, HTTP Aware Network SecurityLinux Native, HTTP Aware Network Security
Linux Native, HTTP Aware Network Security
 
eBPF - Observability In Deep
eBPF - Observability In DeepeBPF - Observability In Deep
eBPF - Observability In Deep
 
Aci presentation
Aci presentationAci presentation
Aci presentation
 
eBPF Basics
eBPF BasicseBPF Basics
eBPF Basics
 
A comparison of segment routing data-plane encodings
A comparison of segment routing data-plane encodingsA comparison of segment routing data-plane encodings
A comparison of segment routing data-plane encodings
 
ACI Hands-on Lab
ACI Hands-on LabACI Hands-on Lab
ACI Hands-on Lab
 
Open ebs 101
Open ebs 101Open ebs 101
Open ebs 101
 
Enable DPDK and SR-IOV for containerized virtual network functions with zun
Enable DPDK and SR-IOV for containerized virtual network functions with zunEnable DPDK and SR-IOV for containerized virtual network functions with zun
Enable DPDK and SR-IOV for containerized virtual network functions with zun
 
cilium-public.pdf
cilium-public.pdfcilium-public.pdf
cilium-public.pdf
 
IPMI is dead, Long live Redfish
IPMI is dead, Long live RedfishIPMI is dead, Long live Redfish
IPMI is dead, Long live Redfish
 
Kubernetes in Hybrid Environments with Submariner
Kubernetes in Hybrid Environments with SubmarinerKubernetes in Hybrid Environments with Submariner
Kubernetes in Hybrid Environments with Submariner
 
The TCP/IP Stack in the Linux Kernel
The TCP/IP Stack in the Linux KernelThe TCP/IP Stack in the Linux Kernel
The TCP/IP Stack in the Linux Kernel
 
Vpc notes
Vpc notesVpc notes
Vpc notes
 

Similar to Introduction to OFI

Middleware Technologies ppt
Middleware Technologies pptMiddleware Technologies ppt
Middleware Technologies ppt
OECLIB Odisha Electronics Control Library
 
Internetworking
InternetworkingInternetworking
Internetworking
Raghu nath
 
Module1 Mobile Computing Architecture
Module1 Mobile Computing ArchitectureModule1 Mobile Computing Architecture
Module1 Mobile Computing Architecture
raksharao
 
Task communication
Task communicationTask communication
Task communication
1jayanti
 
Chapter 3-Processes.ppt
Chapter 3-Processes.pptChapter 3-Processes.ppt
Chapter 3-Processes.ppt
sirajmohammed35
 
Internet basics and Cloud Computing- Manish Jha
Internet basics and Cloud Computing- Manish JhaInternet basics and Cloud Computing- Manish Jha
Internet basics and Cloud Computing- Manish Jha
manish jha
 
Manish Jha- Research Scholar- Internet Basics Requriement
Manish Jha- Research Scholar- Internet Basics RequriementManish Jha- Research Scholar- Internet Basics Requriement
Manish Jha- Research Scholar- Internet Basics Requriement
Manish Jha
 
OSI Model and TCP/IP Model.
OSI Model and TCP/IP Model.OSI Model and TCP/IP Model.
OSI Model and TCP/IP Model.
City University, Dhaka, Bangladesh
 
Osi model and tcpip model.
Osi model and tcpip model.Osi model and tcpip model.
Osi model and tcpip model.
kona paul
 
CN unit 1 part 2 2023.ppt
CN unit 1 part 2 2023.pptCN unit 1 part 2 2023.ppt
CN unit 1 part 2 2023.ppt
mohanravi1986
 
OSI and TCP/IP reference models in networking
OSI and TCP/IP reference models in networkingOSI and TCP/IP reference models in networking
OSI and TCP/IP reference models in networking
MeenakshiGupta233101
 
Assignment3
Assignment3Assignment3
Assignment3
MahfujurRahman63
 
Inter-Process Communication (IPC) techniques on Mac OS X
Inter-Process Communication (IPC) techniques on Mac OS XInter-Process Communication (IPC) techniques on Mac OS X
Inter-Process Communication (IPC) techniques on Mac OS X
HEM DUTT
 
Chapter 4 communication2
Chapter 4 communication2Chapter 4 communication2
Chapter 4 communication2
DBU
 
OPC PPT
OPC PPTOPC PPT
OPC PPT
Shivam Singh
 
SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...
SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...
SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...
ijcsit
 
SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...
SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...
SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...
AIRCC Publishing Corporation
 
Routers BY RAJIT(9816115641)
Routers BY RAJIT(9816115641)Routers BY RAJIT(9816115641)
Routers BY RAJIT(9816115641)
C.U
 
Cataloging Of Sessions in Genuine Traffic by Packet Size Distribution and Ses...
Cataloging Of Sessions in Genuine Traffic by Packet Size Distribution and Ses...Cataloging Of Sessions in Genuine Traffic by Packet Size Distribution and Ses...
Cataloging Of Sessions in Genuine Traffic by Packet Size Distribution and Ses...
IOSR Journals
 
Internet
InternetInternet
Internet
guest076187c
 

Similar to Introduction to OFI (20)

Middleware Technologies ppt
Middleware Technologies pptMiddleware Technologies ppt
Middleware Technologies ppt
 
Internetworking
InternetworkingInternetworking
Internetworking
 
Module1 Mobile Computing Architecture
Module1 Mobile Computing ArchitectureModule1 Mobile Computing Architecture
Module1 Mobile Computing Architecture
 
Task communication
Task communicationTask communication
Task communication
 
Chapter 3-Processes.ppt
Chapter 3-Processes.pptChapter 3-Processes.ppt
Chapter 3-Processes.ppt
 
Internet basics and Cloud Computing- Manish Jha
Internet basics and Cloud Computing- Manish JhaInternet basics and Cloud Computing- Manish Jha
Internet basics and Cloud Computing- Manish Jha
 
Manish Jha- Research Scholar- Internet Basics Requriement
Manish Jha- Research Scholar- Internet Basics RequriementManish Jha- Research Scholar- Internet Basics Requriement
Manish Jha- Research Scholar- Internet Basics Requriement
 
OSI Model and TCP/IP Model.
OSI Model and TCP/IP Model.OSI Model and TCP/IP Model.
OSI Model and TCP/IP Model.
 
Osi model and tcpip model.
Osi model and tcpip model.Osi model and tcpip model.
Osi model and tcpip model.
 
CN unit 1 part 2 2023.ppt
CN unit 1 part 2 2023.pptCN unit 1 part 2 2023.ppt
CN unit 1 part 2 2023.ppt
 
OSI and TCP/IP reference models in networking
OSI and TCP/IP reference models in networkingOSI and TCP/IP reference models in networking
OSI and TCP/IP reference models in networking
 
Assignment3
Assignment3Assignment3
Assignment3
 
Inter-Process Communication (IPC) techniques on Mac OS X
Inter-Process Communication (IPC) techniques on Mac OS XInter-Process Communication (IPC) techniques on Mac OS X
Inter-Process Communication (IPC) techniques on Mac OS X
 
Chapter 4 communication2
Chapter 4 communication2Chapter 4 communication2
Chapter 4 communication2
 
OPC PPT
OPC PPTOPC PPT
OPC PPT
 
SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...
SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...
SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...
 
SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...
SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...
SIMULATION OF SOFTWARE DEFINED NETWORKS WITH OPEN NETWORK OPERATING SYSTEM AN...
 
Routers BY RAJIT(9816115641)
Routers BY RAJIT(9816115641)Routers BY RAJIT(9816115641)
Routers BY RAJIT(9816115641)
 
Cataloging Of Sessions in Genuine Traffic by Packet Size Distribution and Ses...
Cataloging Of Sessions in Genuine Traffic by Packet Size Distribution and Ses...Cataloging Of Sessions in Genuine Traffic by Packet Size Distribution and Ses...
Cataloging Of Sessions in Genuine Traffic by Packet Size Distribution and Ses...
 
Internet
InternetInternet
Internet
 

Recently uploaded

Microservice Teams - How the cloud changes the way we work
Microservice Teams - How the cloud changes the way we workMicroservice Teams - How the cloud changes the way we work
Microservice Teams - How the cloud changes the way we work
Sven Peters
 
GreenCode-A-VSCode-Plugin--Dario-Jurisic
GreenCode-A-VSCode-Plugin--Dario-JurisicGreenCode-A-VSCode-Plugin--Dario-Jurisic
GreenCode-A-VSCode-Plugin--Dario-Jurisic
Green Software Development
 
Everything You Need to Know About X-Sign: The eSign Functionality of XfilesPr...
Everything You Need to Know About X-Sign: The eSign Functionality of XfilesPr...Everything You Need to Know About X-Sign: The eSign Functionality of XfilesPr...
Everything You Need to Know About X-Sign: The eSign Functionality of XfilesPr...
XfilesPro
 
Need for Speed: Removing speed bumps from your Symfony projects ⚡️
Need for Speed: Removing speed bumps from your Symfony projects ⚡️Need for Speed: Removing speed bumps from your Symfony projects ⚡️
Need for Speed: Removing speed bumps from your Symfony projects ⚡️
Łukasz Chruściel
 
Oracle 23c New Features For DBAs and Developers.pptx
Oracle 23c New Features For DBAs and Developers.pptxOracle 23c New Features For DBAs and Developers.pptx
Oracle 23c New Features For DBAs and Developers.pptx
Remote DBA Services
 
WWDC 2024 Keynote Review: For CocoaCoders Austin
WWDC 2024 Keynote Review: For CocoaCoders AustinWWDC 2024 Keynote Review: For CocoaCoders Austin
WWDC 2024 Keynote Review: For CocoaCoders Austin
Patrick Weigel
 
SQL Accounting Software Brochure Malaysia
SQL Accounting Software Brochure MalaysiaSQL Accounting Software Brochure Malaysia
SQL Accounting Software Brochure Malaysia
GohKiangHock
 
Enums On Steroids - let's look at sealed classes !
Enums On Steroids - let's look at sealed classes !Enums On Steroids - let's look at sealed classes !
Enums On Steroids - let's look at sealed classes !
Marcin Chrost
 
Top Benefits of Using Salesforce Healthcare CRM for Patient Management.pdf
Top Benefits of Using Salesforce Healthcare CRM for Patient Management.pdfTop Benefits of Using Salesforce Healthcare CRM for Patient Management.pdf
Top Benefits of Using Salesforce Healthcare CRM for Patient Management.pdf
VALiNTRY360
 
Artificia Intellicence and XPath Extension Functions
Artificia Intellicence and XPath Extension FunctionsArtificia Intellicence and XPath Extension Functions
Artificia Intellicence and XPath Extension Functions
Octavian Nadolu
 
Mobile App Development Company In Noida | Drona Infotech
Mobile App Development Company In Noida | Drona InfotechMobile App Development Company In Noida | Drona Infotech
Mobile App Development Company In Noida | Drona Infotech
Drona Infotech
 
Oracle Database 19c New Features for DBAs and Developers.pptx
Oracle Database 19c New Features for DBAs and Developers.pptxOracle Database 19c New Features for DBAs and Developers.pptx
Oracle Database 19c New Features for DBAs and Developers.pptx
Remote DBA Services
 
一比一原版(USF毕业证)旧金山大学毕业证如何办理
一比一原版(USF毕业证)旧金山大学毕业证如何办理一比一原版(USF毕业证)旧金山大学毕业证如何办理
一比一原版(USF毕业证)旧金山大学毕业证如何办理
dakas1
 
E-Invoicing Implementation: A Step-by-Step Guide for Saudi Arabian Companies
E-Invoicing Implementation: A Step-by-Step Guide for Saudi Arabian CompaniesE-Invoicing Implementation: A Step-by-Step Guide for Saudi Arabian Companies
E-Invoicing Implementation: A Step-by-Step Guide for Saudi Arabian Companies
Quickdice ERP
 
UI5con 2024 - Bring Your Own Design System
UI5con 2024 - Bring Your Own Design SystemUI5con 2024 - Bring Your Own Design System
UI5con 2024 - Bring Your Own Design System
Peter Muessig
 
Using Query Store in Azure PostgreSQL to Understand Query Performance
Using Query Store in Azure PostgreSQL to Understand Query PerformanceUsing Query Store in Azure PostgreSQL to Understand Query Performance
Using Query Store in Azure PostgreSQL to Understand Query Performance
Grant Fritchey
 
Odoo ERP Vs. Traditional ERP Systems – A Comparative Analysis
Odoo ERP Vs. Traditional ERP Systems – A Comparative AnalysisOdoo ERP Vs. Traditional ERP Systems – A Comparative Analysis
Odoo ERP Vs. Traditional ERP Systems – A Comparative Analysis
Envertis Software Solutions
 
E-commerce Development Services- Hornet Dynamics
E-commerce Development Services- Hornet DynamicsE-commerce Development Services- Hornet Dynamics
E-commerce Development Services- Hornet Dynamics
Hornet Dynamics
 
All you need to know about Spring Boot and GraalVM
All you need to know about Spring Boot and GraalVMAll you need to know about Spring Boot and GraalVM
All you need to know about Spring Boot and GraalVM
Alina Yurenko
 
Top 9 Trends in Cybersecurity for 2024.pptx
Top 9 Trends in Cybersecurity for 2024.pptxTop 9 Trends in Cybersecurity for 2024.pptx
Top 9 Trends in Cybersecurity for 2024.pptx
devvsandy
 

Recently uploaded (20)

Microservice Teams - How the cloud changes the way we work
Microservice Teams - How the cloud changes the way we workMicroservice Teams - How the cloud changes the way we work
Microservice Teams - How the cloud changes the way we work
 
GreenCode-A-VSCode-Plugin--Dario-Jurisic
GreenCode-A-VSCode-Plugin--Dario-JurisicGreenCode-A-VSCode-Plugin--Dario-Jurisic
GreenCode-A-VSCode-Plugin--Dario-Jurisic
 
Everything You Need to Know About X-Sign: The eSign Functionality of XfilesPr...
Everything You Need to Know About X-Sign: The eSign Functionality of XfilesPr...Everything You Need to Know About X-Sign: The eSign Functionality of XfilesPr...
Everything You Need to Know About X-Sign: The eSign Functionality of XfilesPr...
 
Need for Speed: Removing speed bumps from your Symfony projects ⚡️
Need for Speed: Removing speed bumps from your Symfony projects ⚡️Need for Speed: Removing speed bumps from your Symfony projects ⚡️
Need for Speed: Removing speed bumps from your Symfony projects ⚡️
 
Oracle 23c New Features For DBAs and Developers.pptx
Oracle 23c New Features For DBAs and Developers.pptxOracle 23c New Features For DBAs and Developers.pptx
Oracle 23c New Features For DBAs and Developers.pptx
 
WWDC 2024 Keynote Review: For CocoaCoders Austin
WWDC 2024 Keynote Review: For CocoaCoders AustinWWDC 2024 Keynote Review: For CocoaCoders Austin
WWDC 2024 Keynote Review: For CocoaCoders Austin
 
SQL Accounting Software Brochure Malaysia
SQL Accounting Software Brochure MalaysiaSQL Accounting Software Brochure Malaysia
SQL Accounting Software Brochure Malaysia
 
Enums On Steroids - let's look at sealed classes !
Enums On Steroids - let's look at sealed classes !Enums On Steroids - let's look at sealed classes !
Enums On Steroids - let's look at sealed classes !
 
Top Benefits of Using Salesforce Healthcare CRM for Patient Management.pdf
Top Benefits of Using Salesforce Healthcare CRM for Patient Management.pdfTop Benefits of Using Salesforce Healthcare CRM for Patient Management.pdf
Top Benefits of Using Salesforce Healthcare CRM for Patient Management.pdf
 
Artificia Intellicence and XPath Extension Functions
Artificia Intellicence and XPath Extension FunctionsArtificia Intellicence and XPath Extension Functions
Artificia Intellicence and XPath Extension Functions
 
Mobile App Development Company In Noida | Drona Infotech
Mobile App Development Company In Noida | Drona InfotechMobile App Development Company In Noida | Drona Infotech
Mobile App Development Company In Noida | Drona Infotech
 
Oracle Database 19c New Features for DBAs and Developers.pptx
Oracle Database 19c New Features for DBAs and Developers.pptxOracle Database 19c New Features for DBAs and Developers.pptx
Oracle Database 19c New Features for DBAs and Developers.pptx
 
一比一原版(USF毕业证)旧金山大学毕业证如何办理
一比一原版(USF毕业证)旧金山大学毕业证如何办理一比一原版(USF毕业证)旧金山大学毕业证如何办理
一比一原版(USF毕业证)旧金山大学毕业证如何办理
 
E-Invoicing Implementation: A Step-by-Step Guide for Saudi Arabian Companies
E-Invoicing Implementation: A Step-by-Step Guide for Saudi Arabian CompaniesE-Invoicing Implementation: A Step-by-Step Guide for Saudi Arabian Companies
E-Invoicing Implementation: A Step-by-Step Guide for Saudi Arabian Companies
 
UI5con 2024 - Bring Your Own Design System
UI5con 2024 - Bring Your Own Design SystemUI5con 2024 - Bring Your Own Design System
UI5con 2024 - Bring Your Own Design System
 
Using Query Store in Azure PostgreSQL to Understand Query Performance
Using Query Store in Azure PostgreSQL to Understand Query PerformanceUsing Query Store in Azure PostgreSQL to Understand Query Performance
Using Query Store in Azure PostgreSQL to Understand Query Performance
 
Odoo ERP Vs. Traditional ERP Systems – A Comparative Analysis
Odoo ERP Vs. Traditional ERP Systems – A Comparative AnalysisOdoo ERP Vs. Traditional ERP Systems – A Comparative Analysis
Odoo ERP Vs. Traditional ERP Systems – A Comparative Analysis
 
E-commerce Development Services- Hornet Dynamics
E-commerce Development Services- Hornet DynamicsE-commerce Development Services- Hornet Dynamics
E-commerce Development Services- Hornet Dynamics
 
All you need to know about Spring Boot and GraalVM
All you need to know about Spring Boot and GraalVMAll you need to know about Spring Boot and GraalVM
All you need to know about Spring Boot and GraalVM
 
Top 9 Trends in Cybersecurity for 2024.pptx
Top 9 Trends in Cybersecurity for 2024.pptxTop 9 Trends in Cybersecurity for 2024.pptx
Top 9 Trends in Cybersecurity for 2024.pptx
 

Introduction to OFI

  • 1. An Introduction to Open Fabric Interfaces Abstract: OFI are high-performance software interfaces that provide low-latency access to fabric hardware. Architectural Overview Open Fabric Interfaces, or OFI, are a set of application programming interfaces (APIs) that export communication services to applications. OFI is specifically designed to meet the performance and scalability requirements of high-performance computing (HPC) applications, such as MPI, SHMEM, PGAS, DBMS, and enterprise applications, running in a tightly coupled network environment. OFI is agnostic to the underlying networking protocols, as well as the implementation of the networking devices. OFI is instantiated in the libfabric library. Libfabric is supported on commonly available Linux based distributions. OFI is architected to support process direct I/O. Process direct I/O, historically referred to as RDMA, allows an application to access network resources without operating system interventions. Data transfers can occur between networking hardware and application memory with minimal software overhead. Although OFI supports process direct I/O, it does not mandate any implementation or require operating system bypass. The following diagram highlights the general architecture of the interfaces exposed by OFI. For reference, the diagram shows OFI in reference to a NIC. This is provided as an example only of how process direct I/O may be supported.
  • 2. OFI Architecture OFI is divided into two separate components. The main component is the OFI framework, which defines the interfaces that applications use. The OFI frameworks provides some generic services; however, the bulk of the OFI implementation resides in the providers. Providers plug into the framework and supply access to fabric hardware and services. Providers are often associated with a specific hardware device or NIC. Because of the structure of the OFI framework, applications access the provider implementation directly for most operations, in order to ensure the lowest possible software latencies. As captured in the architecture diagram, OFI can be grouped into four main collections of interface sets. Control Services: These are used by applications to discover information about the types of communication services are available in the system. For example, discovery will indicate what fabrics are reachable from the local node, and what sort of communication each provides. Communication Services: These interfaces are used to setup communication between nodes. It includes calls to establish connections (connection management), as well as functionality used to address connectionless endpoints (address vectors). Completion Services: OFI exports asynchronous interfaces. Completion services are used to report the results of submitted operations. Completions may be reported using event queues, which provide details about the operation that completed. Or, completions may be reported using lower-impact counters that simply return the number of operations that have completed.
  • 3. Data Transfer Services: These are sets of interfaces designed around different communication paradigms. Although shown outside the data transfer services, triggered operations are strongly related to the data transfer operations. There are four basic data transfer interface sets. Message queues expose the ability to send and receive data with message boundaries being maintained. Message queues act as FIFOs, with sent messages matched with receive buffers in the order that messages are received. Tag matching is similar to message queues in that it maintains message boundaries. Tag matching differs from message queues in that received messages are directed into buffers based on small steering tags that are carried in the sent message. RMA stands for remote memory access. RMA transfers allow an application to write data directly into a specific memory location in a target process, or to read memory from a specific address at the target process and return the data into a local buffer. Atomic operations are similar to RMA transfers, in that they allow direct access to the memory on the target process. Atomic operations allow for manipulation of the memory, such as incrementing the value found in memory. Because RMA and atomic operations provide direct access to a process’s memory buffers, additional security synchronization is needed. Object Model Interfaces exposed by OFI are associated with different objects. The following diagram shows a high- level view of the parent-child relationships. OFI Parent-Child Object Relationships Fabric: A fabric represents a collection of hardware and software resources that access a single physical or virtual network. For example, a fabric may be a single network subnet. All network ports on a system that can communicate with each other through the fabric belong to the same fabric domain. A fabric shares network addresses and can span multiple providers.
  • 4. Domain: A domain represents a logical connection into a fabric. For example, a domain may map to a physical or virtual NIC. A domain defines the boundary within which fabric resources may be associated. Each domain belongs to a single fabric. Passive Endpoint: Passive endpoints are used by connection-oriented protocols to listen for incoming connection requests. Passive endpoints often map to software constructs and may span multiple domains. Event Queues: EQs are used to collect and report the completion of asynchronous operations and events. Event queues handle events that are not directly associated with data transfer operations, referred to as control events. For example, connection requests and asynchronous errors that are not associated with a specific data transfer are reported using event queues. Wait Sets: Although OFI allows the use of native operating system constructs for applications to use when waiting for events, it defines an optimized method for applications to use when waiting on events across multiple event queues, completion queues, and counters. Wait sets allow a single underlying wait object to be signaled whenever a specified condition occurs on an associated event queue, completion queue, or counter. Active Endpoint: Active endpoints are data transfer communication portals. Active endpoints are used to perform data transfers, and are conceptually similar to a socket. Active endpoints are often associated with a single hardware NIC. Completion Queue: Completion queues are high-performance queues used to report the completion of data transfer operations. Unlike fabric event queues, completion queues are often associated with a single hardware NIC. Completion Counter: Completion counters are used to report the number of completed data transfer operations. Completion counters are considered lighter weight than completion queues, in that a completion simply increments a counter, rather than placing an entry into a queue. Poll Set: OFI allows providers to use an application’s thread to process asynchronous requests. This can provide performance advantages for providers that use software to progress the state of a data transfer. Poll sets allow an application to group together multiple objects, such that progress can be driven across all associated data transfers. In general, poll sets are used to simplify applications where manual progress model is employed. Memory Region: Memory regions describe application’s local memory buffers. In order for fabric resources to access application memory, the application must first grant permission to the fabric provider by constructing a memory region. Memory regions are required for specific types of data transfer operations, such as RMA and atomic operations. Address Vectors: Address vectors are used by connectionless endpoints. They map higher level addresses, such as IP addresses, which may be more natural for an application to use, into fabric specific addresses. The use of address vectors allows providers to reduce the amount of memory required to maintain large address look-up tables, and eliminate expensive address resolution and look-up methods during data transfer operations.
  • 5. Connection-Oriented Communication OFI supports both connected and connectionless communication. The following diagram highlights the general usage behind connection-oriented communication. Connected Endpoints Connections require the use of both passive and active endpoints. In order to establish a connection, an application must first create a passive endpoint and associate it with an event queue. The event queue will be used to report the connection management events. The application then calls listen on the passive endpoint. A single passive endpoint can be used to form multiple connections. The connecting peer allocates an active endpoint, which is also associated with an event queue. Connect is called on the active endpoint, which results in sending a connection request (CONNREQ) message to the passive endpoint. The CONNREQ event is inserted into the passive endpoint’s event queue, where the listening application can process it. Upon processing the CONNREQ, the listening application will allocate an active endpoint to use with the connection. The active endpoint is bound with an event queue. Although the diagram shows the use of a separate event queue, the active endpoint may use the same event queue as used by the passive endpoint. Accept is called on the active endpoint to finish forming the connection. It should be noted that the OFI accept call is different than the accept call used by sockets. The differences result from OFI supporting process direct I/O. OFI does not define the connection establishment protocol, but does support a traditional three-way handshake used by many technologies. After calling accept, a response is sent to the connecting active endpoint. That response generates a CONNECTED event on the remote event queue. If a three-way handshake is used, the remote endpoint will generate an acknowledgement message that will generate a CONNECTED event for the accepting endpoint. Regardless of the connection protocol, both the active
  • 6. and passive sides of the connection will receive a CONNECTED event that signals that the connection has been established. Connectionless Communications Connectionless communication allows data transfers between active endpoints without going through a connection setup process. The diagram below shows the basic components needed to setup connectionless communication. Connectionless Endpoints OFI requires the addresses of peer endpoints be inserted into a local addressing table, or address vector, before data transfers can be initiated against the remote endpoint. Address vectors abstract fabric specific addressing requirements and avoid long queuing delays on data transfers when address resolution is needed. For example, IP addresses may need to be resolved into Ethernet MAC addresses. Address vectors allow this resolution to occur during application initialization time. OFI does not define how an address vector be implemented, only its conceptual model. Address vectors may be associated with an event queue. After an address is inserted into an address vector and the fabric specific details have been resolved, a completion event is generated on the event queue. Data transfer operations against that address are then permissible on active endpoints that are associated with the address vector. Connectionless endpoints must be associated with an address vector. Endpoints Endpoints represent communication portals, and all data transfer operations are initiated on endpoints. OFI defines the conceptual model for how endpoints are exposed to applications, as demonstrated in the diagrams below.
  • 7. Basic Endpoint Architecture Endpoints are usually associated with a transmit context and a receive context. Transmit and receive contexts are often implemented using hardware queues that are mapped directly into the process’s address space, though OFI does not require this implementation. Although not shown, an endpoint may be configured only to transmit or receive data. Data transfer requests are converted by the underlying provider into commands that are inserted into transmit and/or receive contexts. Endpoints are also associated with completion queues. Completion queues are used to report the completion of asynchronous data transfer operations. An endpoint may direct completed transmit and receive operations to separate completion queues, or the same queue (not shown). A more advanced usage model of endpoints that allows for resource sharing is shown below. Shared Transmit and Receive Contexts Because transmit and receive contexts may be associated with limited hardware resources, OFI defines mechanisms for sharing contexts among multiple endpoints. The diagram above shows two endpoints each sharing transmit and receive contexts. However, endpoints may share only the transmit context or only the receive context or neither. Shared contexts allow an application or resource manager to prioritize where resources are allocated and how shared hardware resources should be used. Completions are still associated with the endpoints, with each endpoint being associated with their own completion queue(s). The final endpoint model is known as a scalable endpoint. Scalable endpoints allow a single endpoint to take advantage of multiple underlying hardware resources.
  • 8. Scalable Endpoints Scalable endpoints have multiple transmit and/or receive contexts. Applications can direct data transfers to use a specific context, or the provider can select which context to use. Each context may be associated with its own completion queue. Scalable contexts allow applications to separate resources to avoid thread synchronization or data ordering restrictions.