Nutanix Basics ============== Topics to be discussed: NX Node Naming Convention +++++++++++++++++++++++++ *Objective: familiarity with native NX* .. image:: node-naming.png :width: 700 :alt: Nutanix model breakdown Cluster & Node architecture +++++++++++++++++++++++++++ *Objective: prepare audience to be able to draw up our architecture* .. image:: hl-architecture.png :width: 700 :alt: Nutanix High Level Architecture I/O CDP +++++++ *Objective: explain why the architecture delivers high performance I/O and its failsafe data protection* - Reads and writes in an RF2 example - OpLog - Quantity of SSDs used by the OpLog - Software RAID 1 - < 16kb & > 64kb write pathing - Guarantee of how a write operation is committed correctly and a read request is 100% true of its original. Expand on this further to drive pre-fail notification and subsequent offline-ing of a disk - North/South & East/West data locality of virtual machine *hot* data (explain that no other HCI vendor functions in this manner) Data locality +++++++++++++ *Objective: re-iterate the North/South & East/West CDP and where it isn’t a good fit* - Use the example of a VM migration where the data is local, VM moves and where data now resides in relation to its original - When data locality doesn’t work well (de-dupe, ABS and storage only nodes) Replication Factors (RF2 & RF3) +++++++++++++++++++++++++++++++ *Objective: understand how to identify when to use 2 or 3* - Describe the minimum requirements for RF2 & RF3 and in which scenarios would each apply? - What is the ‘maximum cluster size’ for RF2 before tipping into RF3? (provoke a discussion as no actual number exists) - RF3 can suffer up to 25% performance degradation versus RF2 due to its additional write and metadata updates Storage Pools and Containers ++++++++++++++++++++++++++++ *Objective: describe the methods used by the hypervisors and that minimal is the utopia* - Which hypervisors use what protocols - What is the ‘perfect’ amount of Containers in a single Nutanix Cluster? - Store templates and VMs on the same container Compression / Deduplication / Erasure Coding (EC-X) +++++++++++++++++++++++++++++++++++++++++++++++++++ *Objective: explain the different disk optimisation techniques and their relevance* - Compression - Inline and post process - LZ4 inline - LZ4HC 3 days after - Do not use for encrypted data - Deduplication - Block size is 8KB - Erasure Coding - What it is and it's purpose? - 8k & 4k granularity for EC-X - When to use the methods above in different environments - Don't forget to mention the dynamic behaviours and how no other HCI vendors compare to us Data Protection +++++++++++++++ *Objective: describe the differences of each, functionality gained and lost with each hypervisor and why snapshots are not backups* - Our snapshot process - Asynchronous, Synchronous and Metro Availability - Why Synchronous replication is not Metro - The differences in functionality of these features for each hypervisor - NearSync and its pre-requisites - SSD overhead, minimum SSD size and quantity depending on storage requirement plus CPU cores... - Review this document for the latest `PD Documentation`_. .. _PD Documentation: https://docs.google.com/document/d/1A4Z4m-5_EbmOGx2eX0J-5MvoK9mUIaQrpRY_k-w6s8c/edit - CloudConnect - AWS S3 & Azure Blobs (these require compute resource for the CVM) Networking ++++++++++ *Objective: describe the physical (and logical) network elements for Nutanix including the terminology used* .. image:: nx-networking.png :width: 700 :alt: NX Networking - Physical ports on a node - IPMI and failover IPMI - LAN ports - CVM NIC layout - ‘Storage traffic’ separation - Access and Trunk ports - LACP - Hypervisor and CVM ‘same subnet’ requirement - Support ‘supported’ NIC speeds and refer to the Networking Best Practice Guide (BP2050) Mixing and matching of nodes ++++++++++++++++++++++++++++ *Objective: describe the flexibility and ‘configure to order’ approach with the caveats of mixing* - Considerations - Unable to mix hardware vendors in the same cluster - Multiple clusters each using different hardware vendors is permitted - Mixed cluster with storage heavy nodes, the impact of imbalance - Part population of drives - SSD imbalance across nodes and the potential impact to Curator - Different CPUs in the same Nutanix Cluster from AOS, hypervisor and workload perspective - Mixed node configurations in the same block is not permitted, it's not a technical reason Expansion and Reduction +++++++++++++++++++++++ *Objective: describe the flexibility of cluster explanation and node eviction* - Flexible - Add any node of the same hardware vendor - Remove a node from a cluster for re-purposing - Reclaim licenses - Duration of node eviction Failure scenarios +++++++++++++++++ *Objective: preparing for customer failure questions during demos and workshops* - Node failure - CVM failure - …and rolling upgrade - Disk failure - SSD failure (disks 1&2) - On a single SSD node - Node no longer accessible - One a dual SSD node - CVM reboots node to check metadata - NTP failure - Cluster won’t start - Protection domain snapshots fail Shadow Clones +++++++++++++ *Objective: describe why it exists and also that it’s not just for VDI* - Why have it? - What hypervisors are supported? - What triggers the feature on non-AHV clusters? (The three CVM trigger) - Functionality difference between ESXi / Hyper-V and AHV (Copy of write, no waiting for three CVM trigger) Data-at-rest-Encryption +++++++++++++++++++++++ *Objective: awareness to position the software and hardware options and pre-requisites* - The software and hardware options - Differences in functionality between hypervisors - KMS options, third party and Nutanix native Prism / Prism Central +++++++++++++++++++++ *Objective: overview the Element and Central differences, the fully distributed architecture(s) and resilience* - Prism Element navigation and demonstrate how to deliver a ‘10-15 minute first meeting demo’ - Prism Central and Prism Pro features - Prism Central Scale-Out Nutanix Files +++++++++++++ *Objective: overview the feature and use-case(s)* - What is it and why do we have it? - Hypervisors supported - Protocols supported - Replication options Nutanix Volumes +++++++++++++++ *Objective: overview the feature and possible use-case(s)* - What is it and why do we have it? - Hypervisors supported - Protocols supported - Replication options Hypervisors +++++++++++ *Objective: check familiarity of the hypervisors and supporting eco-system components as well as some basics* - VMware vSphere, vCenter, Intelligent Operations (previously vROps), vRealize Automation - Microsoft Hyper-V, VMM, SCVMM, SCOM - AHV - Background, explain the KVM story - Pricing - HA / DRS / EVC equivalent features and our naming conventions