Virtualization, Cloud Computing, and VRF Fundamentals
Virtualization Fundamentals
- Traditional Servers: Historically, there was a 1-to-1 relationship between physical hardware and the Operating System (OS). This is highly inefficient as CPU, RAM, Storage, and NIC (Network Interface Card) capacity are typically underutilized.
- Virtualization: Breaks the 1-to-1 relationship, allowing multiple OSs to run simultaneously on a single physical server.
- Virtual Machine (VM): An isolated software instance of an OS running on virtualized hardware.
- Hypervisor (Virtual Machine Monitor / VMM): The software layer that manages and allocates physical hardware resources (CPU, RAM) to each VM.
Hypervisor Types
| Feature | Type 1 (Native / Bare-Metal) | Type 2 (Hosted) |
|---|---|---|
| Architecture | Runs directly on physical hardware. | Runs as an application on a Host OS. |
| Examples | VMware ESXi, Microsoft Hyper-V. | VMware Workstation, Oracle VirtualBox. |
| Use Case | Enterprise data centers. | Personal devices, home labs (e.g., Cisco CML). |
| OS Terminology | Guest OS runs directly on the hypervisor. | Host OS runs on hardware; Guest OS runs inside the VM. |
Key Benefits of Virtualization:
- Partitioning: Run multiple OSs on one machine, dividing system resources.
- Isolation: Provides fault isolation; a crash in one VM does not affect others.
- Encapsulation: An entire VM can be saved, moved, and copied as easily as a regular file.
- Hardware Independence: Move any VM to any physical server running a compatible hypervisor.
- Cost & Agility: Reduces CapEx (fewer physical servers) and OpEx (less space, power, cooling). Rapid deployment compared to ordering and racking physical hardware.
Virtual Networks
- Virtual Switch (vSwitch): Provided by the hypervisor to connect VMs to each other and the external network.
- Features: Just like physical switches, vSwitch interfaces can operate as access or trunk ports and use VLANs for Layer 2 separation.
- Uplinks: The vSwitch connects to the physical server’s NIC(s) to communicate with the external network.
- Redundancy: vPC (Virtual Port Channel) can be used to form a port channel to two separate physical switches for redundancy.
Cloud Computing Fundamentals
Defined by the NIST (National Institute of Standards and Technology) in SP 800-145. Cloud computing enables ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources.
5 Essential Characteristics
- On-demand self-service: Consumers can provision computing capabilities (e.g., server time) automatically without human interaction with the provider.
- Broad network access: Services are available over the network (Internet/WAN) and accessed via standard mechanisms by heterogeneous clients (mobile, tablets, laptops).
- Resource pooling: Multi-tenant model serving multiple consumers. Resources are dynamically assigned/reassigned. Provides location independence (customer generally doesn’t know the exact physical location).
- Rapid elasticity: Capabilities scale outward and inward rapidly to commensurate with demand. To the consumer, the pool appears unlimited. (Note: The pool is not actually infinite, it just appears so).
- Measured service: Resource usage is metered, monitored, controlled, and reported. Provides transparency and pay-per-use billing.
3 Service Models
| Model | Definition | Customer Controls | Provider Controls | Examples |
|---|---|---|---|---|
| SaaS (Software as a Service) | Use provider’s applications running on cloud infrastructure. | Limited user-specific app settings. | Everything (Datacenter to Application). | Microsoft Office 365, Google G Suite |
| PaaS (Platform as a Service) | Deploy consumer-created apps using provider’s supported languages/tools. | Deployed applications & app-hosting environment. | Datacenter to Runtime/Middleware. | AWS Lambda, Google App Engine |
| IaaS (Infrastructure as a Service) | Provision processing, storage, networks to run arbitrary software. | OS, storage, deployed apps, limited networking (host firewalls). | Physical datacenter, network, servers, storage. | Amazon EC2, Google Compute Engine |
4 Deployment Models
- Private Cloud: Exclusive use by a single organization. Can be on-premises or off-premises. May be owned/managed by a third party (e.g., AWS hosting a private cloud for the DoD).
- Community Cloud: Exclusive use by a specific community of organizations with shared concerns (e.g., mission, security, compliance). Least common model.
- Public Cloud: Open use by the general public. Exists on the cloud provider’s premises (e.g., AWS, Azure, GCP, OCI).
- Hybrid Cloud: Composition of two or more distinct cloud infrastructures (private, community, public) bound together for data/application portability (e.g., cloud bursting for load balancing).
Common Misconceptions:
- Cloud = Off-premises? Not always. Private clouds can exist on-premises within a company’s own facility.
- Infinite Resources: Cloud resources are not actually infinite; they just appear unlimited to the consumer due to rapid elasticity.
Cloud Benefits & Connectivity
- Benefits: Shift from CapEx to OpEx. Global scale, speed/agility, improved productivity, high reliability (geographic mirroring/backups for disaster recovery).
- Connectivity: Enterprises connect to public clouds via private WAN (e.g., MPLS VPN) or the Internet. If using the Internet, a VPN must be used for security. Redundant connections are highly recommended to avoid single points of failure.
Containers vs. Virtual Machines
- Containers: Lightweight software packages containing an App and all its dependencies (binaries, libraries). One container typically runs one app.
- Container Engine: Software that runs containers on top of a Host OS (usually Linux). Example: Docker Engine.
- Container Orchestrator: Automates deployment, management, and scaling of thousands of containers (e.g., Kubernetes, Docker Swarm). Essential for Microservice Architecture (dividing a monolithic app into hundreds of small services).
VM vs. Container Comparison
| Feature | Virtual Machines (VMs) | Containers |
|---|---|---|
| Architecture | Each VM runs its own Guest OS. | No OS inside; shares the Host OS kernel. |
| Boot Time | Minutes | Milliseconds |
| Size | Tens of Gigabytes (GB) | Tens of Megabytes (MB) |
| Resource Usage | Higher CPU and RAM overhead | Very low overhead |
| Portability | High (requires same hypervisor) | Excellent (runs on almost any container service) |
| Isolation & Security | High (OS-level isolation) | Lower (if Host OS crashes, all containers fail) |
graph TD
subgraph Virtual Machines
HW1[Physical Hardware] --> HYP[Type 1 Hypervisor]
HYP --> VM1[VM 1 + Guest OS + Bins/Libs + App]
HYP --> VM2[VM 2 + Guest OS + Bins/Libs + App]
end
subgraph Containers
HW2[Physical Hardware] --> HOS[Host OS]
HOS --> CE[Container Engine]
CE --> C1[Container 1 + Bins/Libs + App]
CE --> C2[Container 2 + Bins/Libs + App]
end
Virtual Routing and Forwarding (VRF)
- Concept: Divides a single physical router into multiple virtual routers, each with its own separate routing table. Think of it as “VLANs for routers.”
- VRF-lite: VRF without MPLS. Used to isolate customer traffic and allow overlapping IP addresses (e.g., multiple customers using
192.168.1.0/24on the same physical router). - Scope: Applies only to Layer 3 interfaces (Router interfaces, SVIs, Routed Ports on multilayer switches).
- Global Routing Table: Interfaces not assigned to a VRF use the default global routing table. Traffic cannot pass between different VRFs unless VRF Leaking is configured.
VRF Configuration & Verification
Scenario: Service Provider Router (SPR1) connects to Customer 1 and Customer 2. Both use subnet 192.168.1.0/30. Without VRF, configuring 192.168.1.1/30 on a second interface would result in an overlap error.
! 1. Create the VRFs in global configuration mode
SPR1(config)# ip vrf CUSTOMER1
SPR1(config)# ip vrf CUSTOMER2
! 2. Verify VRF creation
SPR1# show ip vrf
! 3. Assign interfaces to VRFs and configure IP addresses
! CRITICAL: Assigning a VRF removes any existing IP address on the interface!
SPR1(config)# interface GigabitEthernet0/0
SPR1(config-if)# ip vrf forwarding CUSTOMER1
SPR1(config-if)# ip address 192.168.1.1 255.255.255.252
SPR1(config)# interface GigabitEthernet0/2
SPR1(config-if)# ip vrf forwarding CUSTOMER2
! Overlapping IP address is now permitted because it is in a different VRF!
SPR1(config-if)# ip address 192.168.1.1 255.255.255.252
! 4. View routing tables
! Shows the default global routing table (will be empty if all interfaces are in VRFs)
SPR1# show ip route
! Shows the routing table for a specific VRF
SPR1# show ip route vrf CUSTOMER1
SPR1# show ip route vrf CUSTOMER2
! 5. Test reachability
! Standard ping uses the global routing table (will fail if interfaces are in VRFs)
SPR1# ping 192.168.1.2
! Specify the VRF to ping within that virtual routing table
SPR1# ping vrf CUSTOMER1 192.168.1.2
SPR1# ping vrf CUSTOMER2 192.168.1.2
Common Misconceptions & Exam Tips:
- VRFs do NOT create separate broadcast domains. Router interfaces are already in separate broadcast domains by default.
- VLANs do NOT create separate MAC address tables. A switch maintains a single MAC address table even when using VLANs.
- Always configure the IP address AFTER assigning the interface to a VRF, otherwise the IP will be deleted by the
ip vrf forwardingcommand. - When pinging from a router where all interfaces are in VRFs, you must specify the VRF name, or the router will use the empty global routing table and fail.
- Note: Type 3 hypervisors do not exist.
Summary of Commands
VRF (Virtual Routing and Forwarding)
! Create a VRF
ip vrf [name]
! View configured VRFs
show ip vrf
! Assign interface to a VRF (Clears existing IP)
interface [type/number]
ip vrf forwarding [name]
ip address [ip] [mask]
! View global routing table
show ip route
! View specific VRF routing table
show ip route vrf [name]
! Ping using a specific VRF
ping vrf [name] [destination_ip]