Sase Arch Guide
Sase Arch Guide
Sase Arch Guide
Cisco Public
Cisco SASE
Architecture Guide
September, 2021
Figure 1.
High level SASE Architecture
In this new paradigm, IT requires a simple and reliable approach to protect and connect with agility. This is forcing a
convergence of network and security functions closer to users and devices, at the edge —and is best delivered as a cloud-based,
as-a-service model called secure access service edge (SASE).
Figure 2.
SASE Capability Overview
Cloud computing services offer convenient, pay-as-you-go models that eliminate costly expenditures and maintenance. Cloud
providers host a choice of infrastructure, platform, and software offerings on-site that the “rent”, giving your organization the
flexibility to turn cloud computing services up and down according to changing requirements. There are three main cloud
computing service options:
● Infrastructure-as-a-Service (IaaS) – In this model, a cloud provider hosts infrastructure components that are
traditionally located in on-premise data centers. With IaaS, your organization can choose when and how you want to
administer workloads, without needing to buy, manage, and support the underlying infrastructure
● Platform-as-a-Service (PaaS) – This model is one layer of abstraction above IaaS. Cloud providers, in addition to
providing infrastructure components, also host and manage operating systems and middleware that your developers
need to create and run applications
● Software-as-a-Service (SaaS) – With SaaS, cloud providers host and manage an entire infrastructure, as well as e nd-user
applications. When your company chooses a SaaS model, you do not need to install anything; your users will be able to
log in and begin immediately using the cloud provider’s application running on their infrastructure
Service edge refers to global point of presence (PoP), IaaS, or colocation facilities where local traffic from branches and
endpoints is secured and forwarded to the appropriate destination without first traveling to data center focal points.
The Networking-as-a-Service (NaaS) model refers to the ability to offer network management as SaaS. Likewise, for Security-as-
a-Service (SECaaS). By delivering security and networking services together from the cloud, organizations will be able to securely
connect any user or device to any application without having to install and maintain the network management and security
infrastructure.
SASE Architecture
Figure 3.
SASE Architecture Components
To overcome these cost and performance issues, many organizations are adopting a more decentralized networking approach
to optimize performance, otherwise known as direct Internet access (DIA). DIA is an architecture component in which certain
Internet-bound traffic or public cloud traffic from the branch can be routed directly to the Internet, thereby bypassin g the
latency of tunneling Internet-bound traffic to a central site. The goal of SASE is to connect users and devices, regardless of
location, to any application across any cloud. A secure automated WAN is used to optimize performance by ensuring the faste st,
most reliable and secure path to the cloud.
Branches
Configuring multiple routers connected to different circuits (for example, an MPLS link and a broadband Internet link) to rou te
network traffic efficiently and optimally can be challenging. Beyond simple load balancing, available bandwidth capacity may go
unused during periods of congestion. For example, your broadband Internet connection may be running slowly during a given
period of time, while your costly MPLS link is relatively uncongested and may actually be able to provide faster Internet
connectivity. The inability to aggregate disparate links means wasted bandwidth capacity and lower employee satisfaction.
Software defined wide area network (SD-WAN) combines and optimizes traditional WAN technologies, such as MPLS and
broadband Internet connections. This allows organizations to efficiently route network traffic to multiple remote branch
locations while providing enhanced monitoring and management capabilities. SD-WAN monitors network traffic across all
available links in real-time and dynamically selects the best route for each data packet traversing the network.
A SASE architecture should have the following characteristics for connecting branch locations:
● Flexible, as a service WAN management for on-premises, cloud, and multitenant environments
● Route traffic across different links (MPLS, Internet, 5G, etc.) based on destination
● Route traffic across different links based on cost
● Aggregate multiple links to provide greater total bandwidth
● Rerouting traffic across an alternate link when a link is congested, unstable, or down
● Prioritizing certain application traffic to ensure quality of service
Figure 5.
Roaming User to SASE Cloud
A SASE architecture should have the following characteristics for connecting roaming users:
● VPN as a service to provide network connectivity to private cloud resources
● VPN-less access, leveraging a Zero Trust Access approach
● DIA for off network roaming
Figure 6.
Home Office to SASE Cloud
A SASE architecture should have the following characteristics for connecting the home office:
Once limited to personal apps that employees downloaded to their smartphones, SaaS apps have now become core business
apps supporting critical business functions in the modern digital workplace. This solution addresses the following business flows
for the modern network:
Capability Groups
Take a look at the key security components that comprise a SASE solution.
DNS Security
Figure 8.
DNS Filtering capability group
DNS resolution is the first step when a user attempts to access a website or other service on the Internet. DNS Security logs and
categorizes DNS activity by type of security threat or web content and the action taken, whether it was blocked or allowed.
It is critical that the DNS filter is underpinned by excellent threat intelligence sources. Threat intelligence itself is not a solution
but is a crucial security architecture component. A threat intelligence platform centralizes the collection of threat data fr om
numerous sources and formats and—most importantly—presents the data in a usable format.
A cloud-based web proxy or SWG provides security functions such as web category filtering, web reputation-based filtering and
Web Application Firewall functions along with real-time inspection of inbound files for malware and other threats. Content
filtering by category or specific uniform resource locators (URLs) is used to block destinations that violate policies or compliance
rules.
Network anti-malware inspects files as they traverse the network, using dynamic threat intelligence to check the disposition of
files before they reach the device. File sandboxing is used to open and inspect untrusted files which could compromise an
endpoint.
Figure 9.
SWG capability group
Figure 10.
CDFW capability group
FWaaS is the cloud-based delivery of firewall functionality to protect non-web Internet traffic. This typically includes enabling
intrusion prevention rules for application-level visibility and control.
CASBs help control and secure the use of SaaS applications. The value of CASBs stems from their capability to give insight into
cloud application usage across cloud platforms and to identify unsanctioned use. CASBs use auto discovery to expose shadow IT,
detecting and reporting on the cloud applications that are in use across the network.
A vital ability of CASB is data loss prevention - the capability to detect and provide alerts when abnormal user activity occurs to
help stop both internal and external threats.
Figure 12.
In line Data Loss Prevention
Atlhough included as part of CASB, DLP warrants its own group. A common CASB deployment is to install out of band and to
provide API based DLP functionality. For increased security, DLP should be implemented as a standalone inline feature of the
SASE security stack to catch sensitive information as it passes through the network. This can then be supplemented with DLP
capabilities built into a CASB.
Zero Trust security takes a “never trust, always verify” approach to security. ZTNA verifies user identities and establishes device
trust before granting access to authorized applications, helping organizations prevent unauthorized access, contain breaches,
and limit an attacker’s lateral movement on your network. ZTNA requires a strong, cloud-based, multi-factor authentication
(MFA) solution that ensures users are verified before granted access to specified resources.
Identity
Common across all use cases, identity is the fundamental component of ZTNA. To allow a user to communicate across the
network, one must ensure the user is who they say they are through mechanisms such as MFA. Clients can include both
managed and unmanaged devices. With managed devices, there is option to install endpoint security software to ensure that
devices connecting to trusted networks have not been compromised. Endpoint security verifies that security patches have been
installed and no harmful applications are running on the endpoint before granting access to the network.
Figure 15.
Security controls on a managed device
Unmanaged devices do not provide that luxury. When using an unmanaged device, such as a personal smartphone or PC, the
user can verify their identity using MFA, however, there is no insight into what services are running on the device. Network
controls must be put in place to limit network access and to detect suspicious traffic patterns.
Users are not the only endpoints connected to the network. Building management systems, an example of how the Internet of
Things (IoT) has brought change to the network, monitor building services such as lighting, heating and air conditioni ng. These
devices are not only absent a user, but many do not have the capability to leverage an 802.1X supplicant or a Certificate. In this
case, posture assessment can be used to control devices as they connect to the network. Typically, the device MAC address is
used to uniquely identity the device, and a profile is built using information such as:
All of which allows us to build control policies and assign identifying tags to the devices traffic as it communicates across the
network.
Figure 17.
Security controls on a device that is not associated with a user persona
The traffic destination also has influence on the security controls that is applied to the traffic. When accessing SaaS appli cations,
not all security capabilities are needed. Although DNS filtering is not necessarily a requirement, it is still important that the DNS
lookup for the SaaS service is secured, so the capability requirement still applies. The biggest exclusion from the list of SASE
capabilities is Web security. Many business-critical applications will actually recommend, and some have the requirement for
not breaking the application, to not proxy the traffic. Since the application is typically trusted, the traffic can bypass pr oxy rules
and rely on identity (user and/or device) validation and identity-based authorization policies to control activity on the
application. CASB capabilities in the SASE cloud safeguard access to SaaS applications and provide DLP services to protect data.
Anti-malware must be enabled to inspect all files that go to and from the application.
The Internet is the most untrusted part of any network and therefore requires the most rigorous inspection. DNS filtering blo cks
malicious and unwanted domains, IP addresses, and cloud applications before a connection is ever established. Web security is
used to proxy all of the web traffic for a greater level of visibility and control. An increasing percentage of web traffic is
encrypted, and attackers are exploiting this to hide malware, hoping to avoid detection. Web security can decrypt either all or
selective TLS/SSL encrypted traffic to allow for proper filtering, inspection, blocking, and auditing.
Figure 19.
Security capabilities for accessing the Internet
Although cloud is becoming widely adopted, SASE must still support the existing corporate applications that reside in private
data centers or public cloud instances. Applications can be designated into two buckets; those that are accessible from the
internet and those who are not. External corporate applications may have the ability to be accessed via a reverse proxy, wher e
client’s credentials are checked before navigating through the security stack for application access. It is expected that the
applications support SSO (SAML 2.0) to communicate identities across the web.
Figure 20.
Security capabilities for accessing corporate applications that can be reached from public Internet
Existing legacy apps, or the ones difficult to re-architect to be made compatible to an SSO or Zero Trust model, can be made
available using VPN. In SASE, VPN as a Service (VPNaaS) should be implemented to connect roaming users and home offices to
on premise applications. VPNaaS removes the need of on-premise infrastructure and becomes an easily accessible connection
that is integrated within the SASE cloud platform.
Managed devices (or unmanaged devices with an installed VPN client) may access the network over an encrypted tunnel as if
they were sitting on the corporate network. While some applications suit a reverse proxy implementation, more sensitive
applications may require the extra layer of protection an IPsec tunnel would provide. Access could, through policy, be limite d to
managed devices, where endpoint software can be installed to protect sensitive information from compromised devices.
Industrial control systems (ICS) are ever more connected to corporate IT networks. The fundamental concept of Zero Trust, who
is connected to the network, works a little differently when the device has no user presence. Technology like MFA or SAML
cannot be used as there are no credentials or persona assigned to the device. The “thing” is strictly just a device. In this building
control use case, we identify the necessary tag to add to the traffic using the device posture as discussed in Figure 16. Once that
tag, or device identity has been added to the traffic then policy can be applied as if it was a user sending data to the clou d. All
data must pass through a firewall (is the device speaking to only those intended), be checked for malware (are reports, really
just reports) and have some form of application visibility (is the device doing its intended function). Users who require access to
the building controls application from outside of the network would follow the recommendations as outlined in the other
business flows, depending on how the application has been installed.
Figure 22.
Security capabilities for a building control system to send data to a public cloud application server
In addition, DNS security has been purposely left out for this specific business flow, as an assumption has been made the device
is communicating with a known public cloud application. DNS security is still a fundamental security capability for IoT devic es to
ensure the device only speaks to its intended destination. A typical next step for a compromised device is to make a command-
and-control (C2) callback for control by a remote attacker. This type of attack is stopped using DNS security.
Converge
Security teams are frequently inundated by mountains of data from standalone, point se curity products that do not integrate
with other products and require different knowledge levels and skill sets to operate and maintain. The Enterprise Strategy
Group reports that 31 percent of organizations use over 50 disparate tools, and Cisco research indicates that the majority of
them find it challenging to orchestrate alerts from these different tools. This lack of integration and interoperability make s it
difficult, if not impossible, for security analysts to monitor and correlate security and threat information in real-time.
These challenges have grown exponentially as connected branch and remote offices have proliferated. Each location typically
requires a router and firewall at minimum. In remote and branch locations, these are often purchased as commodity
components that provide limited functionality and remote management capabilities. When switching to DIA at remote
locations, there is a need to deliver the right level of security to users — web security, firewalls, data loss prevention, and so on.
However, it is impractical to buy a separate stack of security appliances for each location. Even if some of these components in
branch locations do include security tools, there are usually no IT personnel in these locations to maintain them. Over time, the
hardware cannot cope with the ever-increasing traffic loads, so the security tools will need to be migrated from these
appliances to the cloud where they can be applied, managed, and audited centrally.
The convergence and orchestration of networking and security into a single pane of glass enables enterprise networking and
security teams to confidently build out their networks with the agility that moder n businesses require. By consolidating secure
access services from a single provider, the overall number of vendors will be reduced, the number of physical and/or virtual
appliances will be reduced, and the number of agents required on an end-user device will be reduced.
Figure 24.
Current Cisco SASE Architecture
Moving to a SASE model will be a gradual process as enterprise IT rethinks how to connect to the distributed information
resources they need. Flexibility will be fundamental as IT chooses among multiple security and networking capabilities that b est
fit their operations, regulatory requirements, and types of applications. Security services can be predominately delivered from
the cloud to provide consistent access policies across all types of endpoints. However, globally distributed organizations may
need to apply security and routing services differently according to regional requirements.
Starting with the endpoints, your chosen method of protection will depend on where your applications are currently hosted.
When protecting applications hosted in the public cloud, or protecting users as they traverse the web, Cisco Umbrella unifies
SWG, DNS, firewall, and CASB functionality in one single integrated cloud-native platform. Built as a micro-services-based
architecture with dozens of points of presence around the world, Umbrella provides the scale and reliability needed to secure
today’s remote workforce and branch networks.
Figure 26.
Cisco Umbrella security capabilities
Finally, Cisco SD-WAN provides an overlay WAN architecture with application optimization to deliver predictable application
performance in multi-cloud environments. Whether it is one site or ten thousand, the Cisco SD-WAN solution leverages an
intuitive, web-based dashboard to give you instant insights about your WAN’s health, access to built-in live tools and packet
capture, and centralized visibility and control over application usage both inside and between your networked sites. Cisco has
two available SD-WAN solutions.
Cisco Meraki’s SD-WAN solution has a globally proven platform that gives enterprises the control to build a SASE solution that
suits their needs today and easily adapts to their needs in the future. Best-in-class networking, network security and endpoint
management are converged on to one platform, in the simplest way imaginable. The platform takes complexity out of every
step of the enterprise SASE journey with open APIs for seamless integration across Cisco technologies and third-party systems.
The Meraki Secure SD-WAN then leverages our industry leading technology and curated subset of the services enable customer
flexibility, and to performantly be protected by edge or cloud security services in a seamless and automatic fashion. In doing so
we connect users to applications and optimize the first mile access from whatever distributed or remote location the customer
is working from. For more information see Cisco SD-WAN powered by Meraki.
Cisco SD-WAN is a secure, cloud-scale architecture that is open, programmable, and scalable. Through the Cisco vManage
console, you can quickly establish an SD-WAN overlay fabric to connect data centers, branches, campuses, and colocation
facilities to improve network speed, security, and efficiency. Comprehensive on-premises and cloud-based security helps
accelerate the transition to a SASE architecture where and when it's needed while increasing user productivity by optimizing
cloud and on-premises application performance with real-time analytics, visibility, and control. For more information see Cisco
SD-WAN powered by Viptela.
The advantage of staying with a single vendor is often these separate products have tight integrations between them as they
move towards a SASE offering. Until a full SASE platform is available, organizations can manage the ir SD-WAN overlay using
dedicated management software and build auto tunnels to Cisco’s Umbrella SIG platform. The SD-WAN overlay can manage
which traffic is sent to the SIG and the SIG platform will manage the security policies of that traffic. As Cisco e xpands capabilities
and provides even more integration of the platforms, adopters of these platforms will gain the benefits and move closer to a
complete SASE solution.
Anti-malware Cisco Advanced Malware Protection (integrated with Umbrella, Firewall &
SD-WAN)
Cisco Threat Grid
Appendix D- References
● Cisco SAFE:
https://www.cisco.com/c/en/us/solutions/enterprise/design-zone-security/landing_safe.html
● Cisco SASE:
https://www.cisco.com/c/en/us/products/security/sase.html
● Cisco SD-WAN powered by Meraki:
https://meraki.cisco.com/sdwhat/en
● Cisco SD-WAN powered by Viptela:
https://www.cisco.com/c/en/us/solutions/enterprise-networks/sd-wan/index.html
● Cisco Umbrella:
https://umbrella.cisco.com/
● SASE for Dummies:
https://umbrella.cisco.com/info/secure-access-service-edge-sase-for-dummies-ebook