2.6 Network Services
Network standards provide DODIIS sites with a consistent, scalable
and reliable approach to support multiple intelligence missions
in a secure network environment. Current DODIIS network standards
are based on the DOD adaptation of the Internet Protocol Suite
(IPS), the DOD Data Communication Protocol Standards (DCPS). These
standards, which identify the options and features of Transmission
Control Protocol/Internet Protocol (TCP/IP) that have been adopted
within DOD, are the standards that are referenced in the TAFIM
and approved for inclusion in DOD Profiles. In Network Service
areas not currently being address by DCPS, the DOD standards are
enhanced with commercial specifications.
The network standards which constitute the DOD recommended protocol
suite have resulted from the Federal Internetworking Requirements
Panel (FIRP) and Profiles for Open Systems Internetworking Technologies
(POSIT) efforts (see section 2.6.2
). The primary source of network specifications are DOD performance
standards which provide the specific DOD options for applicable
base standards defined in Internet Requests for Comments (RFCs).
For convenience a cross reference between the MIL-STDs and base
reference standards is provided in Appendix C
. The DOD performance standards may be obtained from DISA
Center for Standards.
The network service standards cited in this section provide infrastructure
support for the DODIIS Client Server Environment. Network service
standards do not cover all possible standard extensions and variations
for network protocols. Unique requirements and services should
be identified in specific service groups and areas that are impacted
by the requirements. Examples of unique requirements include USMTF
for applications that must process USMTF formatted messages, IP
group multicast which could be used in tactical situations to
reduce network bandwidth in tactical environments, or TACO2 which
is used to disseminate imagery to forward users. Non attribution
in this document does not preclude the use of additional standards
to address specific requirements for interoperability or resource
conservation, but use of unique extensions on DODIIS networks
must be bought to the attention of the DODIIS Management Board
(DMB) and Engineering Review Board (ERB) prior to use on DODIIS
networks.
The base standards that define services for Intelink are now included
in Network standards. Additional guidance and direction for the
Network Information and Discovery and Retrieval (NIDR) standards
can be found in the Intelink Standards Profile.
Network protocols, interfaces, and services cover a broad area
which requires deliberate arrangement of discussions in this section
by protocol layers, based on the Open Systems Interconnect (OSI)
reference model (see table 2-7 ).
Section 2.6.1 discusses network-based
support applications and application level protocol support, including
RPC and Transparent File Access (TFA). Section 2.6.2
identifies requirements for the middle layers of DOD TCP/IP
and discusses interoperability issues concerning the operation
of Open Systems Interconnect (OSI) applications in a TCP/IP network
environment. Section 2.6.3
addresses the physical connection of the DODIIS platform to
the external environment. Topics include LAN architectures and
external communications requirements. Section 2.6.4
briefly discusses network management standards and issues,
and section 2.6.5 covers
security standards and issues. Table 2-8 lists the DODIIS network
services standards.
Table 2-7. OSI Reference Model Layers
Table 2-8. Network Service Standards
2.6.1 Network Applications
This section covers the interfaces and functionality of common
support applications that provide the base network services in
the DODIIS network infrastructure. This section also covers the
application programming interfaces that provide network services
between applications and network resources. This section does
not provide all of the guidance needed to develop consistent access
to network services and additional guidance will be forthcoming
in DODIIS Network Concept of Operations and DODIIS Client
Server Environment Integration Compliance Specification. These
documents will address issues such as:
- Means to track TCP port allocation.
- Guidance for use of Sockets and X Transport Interface (XTI).
- Specific guidance on the construction of RPC calls.
2.6.1.1 Electronic Mail (E-mail) Services
E-mail services define the message formats and services for providing
message exchange between heterogeneous computer systems in a network
environment. Selection of message service standards is directly
tied to selection of directory service standards. Selection of
directory service standards will dictate messaging standards and
vice versa. Messaging standards identify protocols for processing
messages, as well as the proper formats for the messages. The
current installed network base, as well as the eventual transition
to Defense Message System (DMS) standards at each DODIIS site,
should be considered when selecting messaging and directory service
agents.
- X.400.
- E-mail standardization efforts have expanded under OSI, resulting
in the publication of the ITU- TSS standards, X.400 and X.500.
The ITU-TSS X.400 message handling system defines a standard message
format and protocol for the exchange of E-mail, with the additional
benefit of addressing international standards. X.400 allows the
interchange of clear text, facsimile, telex, and other types of
textual information, as well as graphics and voice messages, in
a standard format. The combination of the X.400 and X.500 protocols
fulfills the long-term needs for E-mail interoperability among
heterogeneous networks.
- Defense Message System.
- The DMS consists of all the hardware, software, procedures,
standards, facilities, and personnel used to exchange messages
electronically between organizations and individuals in the DOD.
DMS will provide standards-based messaging by replacing current
organizational (Automatic Digital Network [AUTODIN]), data pattern
traffic, and personal messaging systems throughout the DOD with
products based on X.400 and X.500 standards. DMS Military Messaging
(MM) content builds on X.400-1988 (P48) and Allied Communication
Publication (ACP) 123 to define specific message content known
as P772. In addition, there are specific protocols for Message
Transfer (P1), Message Transfer System (MTS) Access (P3), and
Message Store Access (P7). Message security services, such as
data confidentiality, data integrity, data origin authentication,
and (optionally) non-repudiation with proof of origin (digital
signature), will be offered for both organizational and individual
messages, and for both classified and unclassified messages.
- DMS Standardized Profiles:
- The standardized profile for DMS common messaging is specified
in MIL-STD-2045-17501 which is in five parts. All interpersonal
and messaging implementations that interface with DMS must comply
with the corresponding component of MIL-STD-2045-17501 which include
the following:
- MIL-STD-2045-17501-1, MHS Service Support.
- MIL-STD-2045-17501-2, Specification of ROSE, RTSE, ACSE, Presentation
and Session Protocols for use by DOD MHS.
- MIL-STD-2045-17501-3, Requirements for Message Transfer (P1).
- MIL-STD-2045-17501-4, Messaging Requirements for MTS Access
(P3).
- MIL-STD-2045-17501-5, Messaging Requirements for MS Access
(P7).
In addition, there is a five part set of standardized profiles
that identify specific services that must be provided by Military
Messaging (MM) components of a Message Handling Systems (MHS)
to meet DMS standards including:
- MIL-STD-2045-17502-1, MM MHS Service Support
- MIL-STD-2045-17502-2, MM Content (P772)
- MIL-STD-2045-17502-3, MM Requirements for Message Transfer
(P1)
- MIL-STD-2045-17502-4, MM Requirements for MTS Access (P3)
- MIL-STD-2045-17502-5, MM Requirements for MS Access (P7)
The MHS profiles provide the necessary conformance statements
to promote system interoperability and provide the basis for the
development of uniform, internationally recognized system tests.
The profiles identify whether specified MM services are mandatory,
optional, conditional, out of scope, not applicable, or prohibited.
The profiles also indicate the specific MHS components that are
to implement each MM service.
X.400 defines a variety of security services. However, the DMS
Program has adopted the Message Security Protocol (MSP) as the
target security protection mechanism for all DMS organizational
and individual message traffic. MSP was developed under the auspices
of the Secure Data Network System (SDNS) Program concurrent with
international development of the ITU-TSS X.400 1988 Recommendation.
SDNS MSP and 1988 X.400 offer a similar set of security services,
but the two approaches diverge due to differing priorities and
requirements, and the operational environment of the U.S. Department
of Defense. The divergence introduces incompatibilities between
the two approaches, and the native X.400 protection of messages
is not approved for DODIIS use. DMS MSP requirements are
specified in MIL-STD-2045-18500 which is a five part standard
including:
- MIL-STD-2045-18500-1, MSP Service Support.
- MIL-STD-2045-18500-2, MSP Content Protocol.
- MIL-STD-2045-18500-3, MSP Requirements for Message Transfer
(P1)
- MIL-STD-2045-18500-4, MSP Requirements for MTS Access (P3)
- MIL-STD-2045-18500-5, MSP Requirements for MS Access (P7)
- DMS Transition:
- Transition from current messaging systems baselines to DMS-compliant
systems will be accomplished through the use of gateways to provide
the conversion between DMS and AUTODIN formats. Specific requirements
for message handling transition of record message traffic are
identified in ACP 127. In addition, there is a requirement for
SMTP/X.400 transition gateways compliant with MIL-STD-2045-17503-2
for personal messaging. A complete description of DMS transition
planning is documented in the DMS Target Architecture and Implementation
Strategy, dated January 1993. DIA is drafting a DMS transition
plan that covers transition planning for DODIIS messaging components.
Sites should consult the DIA and DMS transition plans when attempting
to identify future site messaging requirements and transition
planning.
DMS implementation will not affect near term DODIIS E-mail and
Directory services, but long term DODIIS planning must take into
account the impact of DMS architecture on all messaging and directory
services requirements. DMS plans to use COTS products or non-developmental
item (NDI) technology whenever possible. It is important to note
that purchase of commercial products that comply with commercial
X.400 and X.500 standards does not guarantee interoperability
with DMS standards.
The DODIIS E-mail standards, X.400 and X.500, are not fully supported
by the TCP/IP protocols at this time. To adopt OSI applications
within TCP/IP environments requires the installation of a mail
gateway. Sites that are currently configured with TCP/IP Simple
Mail Transfer Protocol (SMTP) and Multimedia Internet Mail Extensions
(MIME) should, at minimum, provide gateways, compliant with MIL-STD-
2045-17503-2, to X.400. New development efforts should use X.400
products and sites with current implementations of SMTP should
plan for the transition to DMS compliant X.400 products. Any new
DODIIS mail products selected should, at a minimum, have
an X.400 gateway that complies with MIL-STD-2045-17503-2 if the
products are not X.400 compliant. It is expected that the adoption
of X.400 E-mail standards will expand upon the capabilities supported
by the DOD TCP/IP mail protocols and maintain a migration path
towards DMS messaging services.
2.6.1.2 Directory Services
Directory Services (DS) allow users to identify, by name, network
resources such as servers, files, disks, print queues, or mail
boxes and gain access to the resources without needing to know
where they are located in a network. This versatility allows a
user to continue to access a resource by one name even when attributes
of that resource, such as the resource address, are changed. To
support directory services for applications, computer networks
require names and directories which describe and record the attributes
of all network resources and summarize the information these resources
provide.
- Domain Name System (DNS).
- DNS defines a distributed way of performing host name/IP address
conversion. In addition to name/address conversion, the DNS may
contain associated resource records which describe the host, its
capabilities in terms of well known services (e.g., TCP, IP, SMTP),
and other information such as E-Mail routing using Mail Exchange
(MX) records. DNS has two parts: a domain name resolver, which
runs on every host in the network that runs DNS; and a domain
name server, which responds to queries from resolvers about specified
hosts. DNS is the current DODIIS standard for directory services.
DNS is a public standard used in numerous public and commercial
networks and is readily available. Implementation of DNS in the
DODIIS network environment requires conformance to MIL-STD-2045-17505.
- DNS provides APIs that allow application programs to use DNS
services. DNS is supportable in DODIIS TCP/IP networks and will
be accessed by some future X.500 products, but DNS should eventually
be phased out during the transition to DMS based standards as
DMS compliant directory service products become available. The
implementation vehicles for transition of DODIIS products from
DNS to X.500 will be defined and coordinated as part of the Joint
Worldwide Intelligence Communications System (JWICS; network)
and Client Server Services (workstation/server) development efforts.
Each program has specific responsibilities to implement components
of directory services.
- The Internet Engineering Task Force (IETF) Domain Name System
Security (DNSSEC) Working Group is working on extending the DNS
to provide data origin authentication and transaction authentications
for DNS operations . The extensions also add resource records
to associate encryption keys with DNS names and, therefore, permit
the DNS to be used as a general public key distribution system
in support of both DNS services and many other security services
in the IPS.
- X.500 Network Directory Service.
- X.500 provides Directory Service standardization, and creates
a uniform addressing scheme for electronic message traffic. X.500
allows users to locate users and resources on networks with a
heterogeneous mix of services and workstations. X.500 is supported
as a strategic part of DMS implementation and is needed to support
X.400 message handling. A DOD X.500 profile for X.400 message
handling systems is to be developed during the fiscal year 1995.
This profile will be based on the 1993 version of ITU-T X.500
or International Standards Organization (ISO) 9594. In the meantime,
X.500 base standards should be used as the current reference.
- Each site should determine whether it is more convenient to
implement X.500 directory services now or to implement DNS now
and replace with X.500 in a few years. Commercial versions of
X.500, based on X/Open Directory Service (XDS) specifications
and compatible with TCP/IP standards, are starting to emerge and
could provide an opportunity to position some DODIIS installation
architectures for future transition to DMS standard products.
If X.500 directory services are implemented early, the X.500 product
should have a gateway for access to DNS services to ensure compatibility
with DNS locations and a smooth transition path.
Message handling application software that requires access to
X.500 directory service functions should develop directory service
APIs according to Institute of Electrical and Electronics Engineers
(IEEE) P1224.2, which is a language independent specification
for standard directory service message handling user agents and
should be tested using the test methods identified in IEEE P1326.2,
Language Independent Test Methods. An alternate API specification
(IEEE P1327) is available for implementation in the C language.
Note that the IEEE P1003.17, which was previously cited as the
DODIIS standard, has been replaced by IEEE P1224.2. DMS specific
directory requirements and test standards are discussed in section 2.6.1.1
.
2.6.1.3 File Transfer Services
File transfer services are a set of agreements on procedures that
specify how information organized into files can be transferred
from one computer to another through homogeneous and heterogeneous
networks. These services provide for the transfer of file data
to another location only and do not address data formats or translations
which must be accomplished by the source and target systems applications.
- File Transfer Protocol (FTP).
- FTP is the TCP/IP-based standard currently used to transfer
files between systems throughout the DODIIS network. FTP is mature
and available on numerous processing platforms. FTP provides the
ability to initiate file transfer from either end of a valid connection:
send (push) a file to a receiving system or receive (pull) from
a distant host. The FTP standard is specified in MIL-STD-2045-17504
and DODIIS FTP implementations must use the minimum implementation
identified in MIL-STD-2045-17504.
2.6.1.4 Remote Login Services
Terminal standards identify services and protocols for remote
terminal access to network computing resources. Although the current
DODIIS transition is towards a common user interface and distributed
computer access of remote processing resources, there will continue
to be a requirement for remote terminal access for selected activities.
MIL-STD-2045-17506, Telnet, identifies standards for a simple
asynchronous terminal capability. DODIIS discourages the use of
Network Virtual Data Entry Terminal (NVDET) configurations except
in those cases where NVDET is required to access a legacy NVDET
host. All NVDET-based hosts are required to transition to TN3270
support as soon as feasible. If the send-location option is required
for a host/terminal identifier, the implementation must use DODIIS
standard user identifiers.
2.6.1.5 Network Information Discovery
and Retrieval Protocols
This section discusses some of the OSI application layer protocols
popular on the Internet, commonly referred to as the Network Information
Discovery and Retrieval (NIDR) protocols. All of the following
protocols are adopted by DODIIS because they are required by Intelink.
Intelink will be an integrated Intelligence Information Service
for the Intelligence Community (DOD and non-DOD) that defines
intelligence information exchange services, types of media to
be exchanged, standards, services required of underlying networks,
responsibilities of providers, and user interface requirements.
Security is now becoming increasingly important for NIDR tools.
Added security capabilities, such as data origin authentication,
data integrity, and data confidentiality, are planned for many
of these tools. The DODIIS community is monitoring NIDR tool security
related activities (e.g., Secure HyperText Transfer Protocol (HTTP)).
- Gopher.
- The Internet Gopher (RFC 1436) is a client server document
search and retrieval protocol. A client connects to a server using
TCP, and sends a one-line text "selector string". The
server responds by returning the item (a file, a directory listing,
or a link to some other service) corresponding to the selector
string and immediately closing the connection. Information in
Gopher is organized and presented as a series of networked hierarchical
directories similar to a familiar file system. However, the links
in the Gopher may define services other than simple files or directories,
including Telnet sessions, links to other Gopher servers, and
links to gateway servers. Because of its simple and connectionless
nature, Gopher servers make minimal demands on their host machines
and Gopher clients are easy to implement.
- There are various public domain and commercial implementations,
including the original Gopher software developed at the University
of Minnesota. Also, Veronica (Very easy rodent-oriented net- wide
index to computerized archives) has been developed at the University
of Nevada, to facilitate search for items by title. A central
server periodically scans the complete menu hierarchies of Gopher
servers and the resulting index is provided by a Veronica server
and can be accessed by any Gopher client.
- Z39.50.
- American National Standards Institute/National Information
Standards Organization (ANSI/NISO) Z39.50-1988, OSI Text/Data
Retrieval Protocol for Library Applications in Client Server Environments
, is a general purpose search and retrieval mechanism that
could be used with a wide variety of data types. It allows databases
on the remote system to be searched using search attribute sets
and record syntax for different types of data. Z39.50 may be used
with or without the presentation protocol, and most of its implementations
currently run directly over TCP/IP.
- Wide-Area Information Server (WAIS), developed by Thinking
Machines Corporation, is a tool that uses Z39.50 and includes
user interfaces for most platforms, and server software that provides
automatic indexing of databases. Both freely available and commercial
versions of WAIS servers and clients are available. The user employs
a WAIS client to attach to a particular WAIS server, and specifies
a search pattern which is matched against the server's index.
In early WAIS clients, searches are specified as simple natural-language
queries. Each WAIS server reads the query and searches the full
text of the database for the most relevant documents, and ranks
them using automatic word weighting. The user may choose to vie
selected documents, or further refine the search.
- HTTP.
- HTTP is an Internet Draft that specifies a fast search and
retrieval protocol used by the World Wide Web (also called WWW,
W3, or the web) to transfer hypertext documents, operating in
a fast and stateless way necessary for hypertext jumps. While
the Internet refers to the physical side of the global network
which is a giant mass of cables and computers, the W3 refers to
a body of information available on the Internet and the hypertext
information systems that provide access to this information. There
is a range of W3 browsers, supporting many environments.
- Although an HTTP specification is available as an Internet
Draft, an HTTP Working Group is just being formed in the IETF.
It may take until 1996 before a draft specification advances on
the standards track. Several incompatible, proprietary security
extensions to HTTP have been proposed, and the new working group
will have to resolve these incompatibilities.
- The default primary W3 document format is HTML+, as HTTP
uses links embedded in HTML+ documents for navigation. HTML+ is
typically used to mean both the document type and the markup language
for representing instances of that document type. W3 uses HTTP
to allow transfer representations to be negotiated between client
and server, the result being returned in an extended MIME message.
HTML+, currently an Internet Draft, is proposed to be registered
as a MIME (RFC 1341) content type. Uniform Resource Locators (URLs)
are used to specify the hypertext links in HTML+ documents.
- HTML+ is not to be confused with the functionality of web
browsers such as "Mosaic". HTML+ applies to a document
instance of the SGML DTD of the same name and therefore refers
to the data that is marked up using the SGML-compliant
HTML+; web browsers are the software that takes the HTML+
data and displays in a specific way (e.g., blue and underlined
for hypertext links). As SGML is typically concerned with the
logical structure of a document without addressing the formatting
structure (i.e., how it should look on screen or paper), the SGML-compliant
HTML+ documents need another standard that address formatting.
Rather than implementing the draft DSSSL or MIL-M-28001 Output
Specification for formatting, most web browsers have hard-coded
these rendering information in the software itself. These capabilities
are specific to the web browsers that use relevant information
in the HTML+ documents. See section 2.4.5.2 for more information
on SGML and DSSSL.
- Prospero.
- The Prospero Protocol, Version 5, from the University
of Southern California, specifies a distributed file system within
which users construct their own virtual file systems by selecting
objects and services available over the network.
- Prospero is layered on top of a reliable delivery mechanism
using User Datagram Protocol (UDP). Although the Prospero protocol
is not an Internet draft, it is widely used on the Internet. For
example, it is used by Archie, a popular tool that provides an
electronic directory service for locating information on Internet,
originally created to track the contents of anonymous FTP archive
sites.
- Uniform Resource Locator (URL).
- URL is an Internet draft standard for specifying an object
on the Internet, where the object can be a file in a directory
on any machine on the network, a query, a document in a database,
or results of a command of an Internet tool. URL is one of three
standards that are part of the Uniform Resource Identifiers (URIs),
which define the encoding of system independent resource location
and identification information for the use of Internet information
services. The other two URI standards are:
- - Uniform Resource Name (URN): an identifier which can be
used to uniquely identify a resource and is designed to provide
persistent naming for networked objects. This name would stay
the same regardless of the object's current location(s).
- - Uniform Resource Characteristics (URCs): a method for encapsulating
meta-information about a given URN or URL.
- Network News Transfer Protocol (NNTP).
- Defined in RFC 977, NNTP is a protocol for the distribution,
inquiry, retrieval and posting of news articles using a reliable
stream-based transmission of news among the Internet community.
The format of News articles is defined in RFC 850, Standard
for Interchange of Usenet Messages. This in turn refers to
RFC 822, Standard for the Format of Advance Research Projects
Agency (ARPA) Internet Text Messages, which defines the format
of Internet mail messages.
- NNTP does not provide security services. However, independent
of the NNTP protocol, individual news articles can be protected
the same way that individual messages are protected in a message
handling system.
2.6.1.6 Remote Procedure Calls
The fundamental component of distributed computing is the Remote
Procedure Call (RPC). RPCs allow applications to invoke procedures
that reside on remote hosts. The input parameters for the procedure
are passed in the request, and the outputs of the called procedure's
execution are returned in the reply. RPCs require UDP for transport
level network communications between hosts (see section 2.6.2.1
). This capability is required for DODIIS systems that need
generalized client server capabilities over a network.
Complete distributed computing environments are being developed
around two of the leading de facto RPC implementations. These
RPCs are Sun Microsystem's RPC and Open Software Foundation' (OSF)
RPC which is adopted from Hewlett-Packard/Apollo's Network Computing
System. The complete distributed computing environments are Sun's
Open Network Computing (ONC+) and OSF's Distributed Computing
Environment (DCE), respectively. The ultimate DODIIS goal is to
shift from proprietary de facto specifications for distributed
computing to formally established standards. However, work is
still underway to provide formal standards for IEEE P1003.1g.
The state of formal standards for distributed products mandates
the continued use of proprietary specifications in the near future.
The 1993 version of the DODIIS Profile of the DOD Technical
Reference Model for Information Management specified Sun ONC+
RPC as a DODIIS standard because of the commercial availability
of Sun products for multiple platforms and ONC+ compatibility
with a large percentage of existing DODIIS systems. Under COSE
agreements among major Unix vendors (see section 2.7.1.3
), Sun Microsystems stated its intentions to support DCE and
interface to DCE-based systems within the context of the ONC+
protocol. The most recent change in the commercial world has been
the broad acceptance of DCE-based products. DCE provides an integrated
distributed computing environment that is positioned to meet a
majority of DODIIS distributed computing needs and is noted as
the migration goal for DODIIS distributed computing.
It has been determined that DODIIS will support Sun ONC RPC only
as a legacy standard for current DODIIS systems. All new DODIIS
development will specify compliance to OSF DCE RPC standards.
The DCE RPC is readily available and is public domain.
The security situation for RPC is mixed:
- Confidentiality. DCE has a packet privacy option. ISO
Remote Operations may run over protocols that provide a confidentiality
service, such as Transport Layer Security Protocol (TLSP) or Network
Layer Security Protocol (NLSP). Sun has no specific provision
for confidentiality within ONC, although it too can run over lower-layer
security.
- Integrity. DCE has a packet integrity option. Neither
ISO Remote Operations nor Sun RPC provide integrity service, but
they can run over a lower-layer integrity service.
- Authentication. DCE supports data origin authentication.
Neither Sun RPC nor ISO Remote Operations explicitly support data
origin authentication, but they can over lower-layer services.
2.6.1.7 Transparent File Access
Standards for sharing and managing data in heterogeneous and homogeneous
networks are still being developed. Within the POSIX standardization
effort, the IEEE P1003.1f committee is in the process of defining
TFA standards. One objective of the TFA standardization effort
is to define a specification for the file system semantics which
is required for file sharing within a distributed environment.
Early indications are that P1003.1f is based on the full file
system implementation identified in the base (P1003.1) kernel
specification.
Until recently, NFS has been the de facto DODIIS distributed
file standard as defined in RFC 1094, NFS: Network File System
Protocol Specification, which is built upon the Sun RPC. Because
NFS allows the transparent sharing of files within a heterogeneous
TCP/IP network, this protocol is currently applicable to many
intelligence community applications. On the other hand, OSF DCE
Distributed File System (DFS) satisfies the vast majority of DODIIS
transparent file access requirements including caching, replication,
and access control through the use of access control lists. The
functionality of both NFS and the OSF DFS are similar, but they
are based on different RPCs and as such are not interoperable.
DCE has implemented a gateway to support NFS file structures in
a DCE environment.
NFS can be used for legacy applications that require NFS support,
but all new DODIIS development must specify compliance with DFS
file standards. DFS is the migration goal for DODIIS client server
environment.
2.6.1.8 Distributed Authentication
The adoption of a standard distributed authentication capability
has been identified as a key security requirement for the DODIIS
community. A distributed authentication capability will allow
DODIIS users to identify themselves only once during initial login
via a unique global identifier and authentication mechanism. Any
subsequent need for identification and authentication (e.g., when
accessing a server) will be handled by the distributed authentication
function and would therefore be transparent to the user. As a
result, a user will not be required to remember the user IDs and
authentication mechanisms (e.g., passwords) for each DODIIS resource.
Each DODIIS resource owner, however, will retain control over
who can access their network resources.
DODIIS intends to transition to systems that support a stronger
authentication service than those offered by traditional password-based
systems. Passwords are vulnerable to interception, replay, spoofing,
and dictionary attacks. A distributed authentication capability
that is based on cryptographic techniques is desired.
DODIIS intends to take advantage of standards-based distributed
authentication technologies. Distributed authentication is an
active work item within standards-setting or standards-approving
bodies. The ISO is actively developing an authentication framework.
The IETF Common Authentication Technology (CAT) Working Group
is currently developing a specification on a unified Internet
authentication mechanism that is a hybrid of MIT Kerberos secret-key
and Digital Equipment Corporation's (DEC's) Distributed Authentication
Security Service (DASS) public-key techniques. Both Kerberos and
DASS employ cryptographic mechanisms to provide strong authentication.
The goal of both Kerberos and DASS is to provide authentication
services in a distributed environment that is more secure and
easier to use and manage than existing mechanisms (e.g., traditional
passwords). A near-term goal for DODIIS is to acquire COTS products
that conform to the IETF CAT specification. The IETF CAT specification,
when completed, will be released as a formal Internet standard.
DODIIS intends to take advantage of standards-based distributed
authentication technologies. Distributed authentication is an
active work item within standards-setting or standards-approving
bodies. The ISO is actively developing an authentication framework.
The IETF CAT Working Group has defined and continues to work on
a Generic Security Service API (GSS-API) that provides authentication
services that are implementable by either secret-key (e.g., Kerberos)
or public-key (e.g., Digital Equipment Corporation's (DEC's) DASS)
techniques. DODIIS will continue to monitor the IETF CAT and GSS-
API activities for potential inclusion within the DODIIS Profile.
DODIIS intends to take advantage of National Security Agency (NSA)
Multilevel Information System Security Initiative (MISSI) technologies
and services for distributed authentication security needs. One
such technology is the Fortezza; crypto card (formerly known as
the Tessera; crypto card). The DODIIS community plans to use the
Fortezza crypto card as a strong authentication mechanism. The
Fortezza crypto card will also provide other security services
such as data integrity, non-repudiation, and confidentiality.
Additional information on MISSI is provided in section 2.6.5.4.
2.6.1.9 Network Time Services
Distributed network systems need a consistent time service. Many
distributed services, such as distributed file systems and authentication
mechanisms compare time stamps from different network clients.
DODIIS must support a consistent standard for time stamps for
any useful comparison. Network Time Services can come from numerous
sources, but a network time standard has been established to eliminate
redundant efforts for providing network clients with unique time
formats. RFC 1305, Network Time Protocol (NTP), Version 3, establishes
the standard format for accessing time services in the DODIIS
network architecture. NTP is accurate to 1 - 50 microseconds.
Simple Network Time Protocol (SNTP), RFC 1361, provides time service
that may not be as accurate, but still uses the NTP time format.
SNTP should be used in all cases except where the higher level
of accuracy in the full NTP standard is required. The DODIIS Client
Server Environment program has integrated NTP as a standard component
of their architecture.
NTP provides several security services on a required or optional
basis. It has cryptographic mechanisms for integrity and authentication,
and it uses statistical averaging techniques to determine the
correct time given a multiplicity of time sources, some of which
may be false. Confidentiality and non-repudiation are ignored.
2.6.1.10 Distributed Computing Environment
(DCE)
DCE provides a standard integrated distributed computing environment
that operates consistently across heterogeneous platforms (see
figure 2-3 ). In addition, DCE provides
the functions needed to satisfy several critical DODIIS infrastructure
requirements. DCE is a tightly integrated environment with DCE
components relying on services from other DCE components and as
such can not be broken up for implementation. DCE components are
compliant with some formal requirements that meet DODIIS migration
goals and provide gap standards in areas where there currently
are no formal standards available. The goals of this discussion
is to identify the basic functional architecture of DCE and relate
DCE standards to current DODIIS standards goals.
Figure 2-3.
DCE Architecture Diagram
DCE functions include:
- Threads - DCE Threads provides the ability to increase
system performance through the use of parallelism. In fact, DCE
Threads can provide improved application performance on single
processor systems by permitting overlap of slower I/O operations
with faster computational operations depending on effective design
of applications to concurrent components.
- RPC - DCE provides a consistent implementation of RPC
across various vendor platforms to increase software interoperability
and portability. The DCE RPC is based on a simple model which
shields the applications programmer from details of network communications
between client and server nodes. Programmers that use the standard
DCE RPC APIs do not need to rewrite applications in order to port
them to different architectures, operating systems, communications
protocols, or languages. DCE RPC provides a high level programming
model to the distributed application programmer, hiding communications
details, and removing non portable system and hardware dependencies.
DCE RPC is also integrated with DCE security services to provide
authentication for distributed programs.
- Authentication - DCE meets the DODIIS requirement for
a single logon using Kerberos technology for distributed authentication.
The Kerberos technology also addresses the additional security
requirements of data integrity and privacy through the use of
encryption methods. Future versions of DCE have indicated that
they will support public key technology, in addition to the private
key (Kerberos) technology they now serve.
- Access Control - DCE provides access control lists
as part of Cell Directory Server (CDS). The OSF DCE access control
lists are a superset of the POSIX 1003.6 Access Control List (ACL)
working group specifications. OSF extended P1003.6 standards in
DCE to make them useful in a distributed computing environment.
- DCE access control lists provide the ability to group multiple
resources together eliminating the need to maintain redundant
access control lists. DCE also allows lists of users that are
to have ownership and administrative rights to the resource. DCE
can even provide the capability for delegation of access control
administration privileges to trusted individuals outside the organization
at the discretion of the resource administrator.
- Directory Services - DCE provides two levels of directory
services. CDS is integrated with DCE security services to provide
local name services, as well as, authentication and access control
for local network resources. This service is scalable and distributed
to local domains established in the network architecture and provides
control and maintenance under a local resource manager. DCE also
provides access to a Global level of directory services, Global
Directory Services (GDS), through a Global Directory Agent (GDA)
which is integrated locally in CDS. The GDA can interface with
DNS or X.500 GDS servers.
- File Services - DCE DFS is an asynchronous file system
that is tightly integrated to DCE security to provide authentication
and access control and to DCE directory services to enforce a
global uniform name space and provide consistent user transparent
access to file resources no matter where they are located on a
global network.
- DFS was engineered to address some of the common technical
issues noted in NFS implementations. DFS allows caching and replication
of files on a global basis which provides for reduced communication
bandwidth requirements. DFS uses tokens to track the status of
files and only issues a write token when a request is made for
file modification. DFS also provides management tools that allow
backup and relocation of files while the server is on-line, as
well as, a log based system that provides a file recovery capability
after system crashes.
- Time Services - DCE provides Distributed Time Services
(DTS) in a hierarchical architecture that is transparent to most
DCE users. Programmers who must use DTS services are provided
with a consistent API for programming DTS access from their developed
applications. DTS also has a time provider interface which allows
DTS to synchronize to NTP standards based external clock sources.
Each DCE host will have at least a DTS clerk to access DTS services.
There can be several time servers for every DCE cell, but there
must be one Global Time Server for each cell. DTS is integrated
with other DCE services that require accurate time services.
DCE is part of the DODIIS goal architecture. The standards based
structure of DCE and the consistent distributed computing interface
across heterogeneous platforms makes DCE a high priority for future
DODIIS migration development efforts. DCE is consistent with objective
DODIIS standards, as shown in table 2-9.
Table 2-9 . Comparison of DCE and
DODIIS Goal Standards
[ TOC ]
[ Back ]
[ Next ]
DoDIIS Profile of the Technical Reference Model - Feb 1995 - Draft