Sql Server Licensing Vmware – VSphere Default Maximum Configuration Configuration Item ESX 5.5 ESX 6.0 Virtual CPUs per virtual machine (Virtual SMP) 64,128 RAM per virtual machine 1 TB 4TB Virtual machine swap file size Virtual NICs per virtual machine 10 16 sockets per host 480 logical per host CPU 320 virtual machines per host 512 2048 virtual processors, 32 virtual processors per FT virtual machine. 1 4 FT virtual machines, RAM per host.
Physical Hardware VMware HCL BIOS / Firmware Power / C-States Hyper-threading NUMA ESXi Host Power Virtual Switches vMotion Portgroups Virtual Machine Resource Dedicated Storage Memory CPU / vNUMA Network vSCSI Controller Guest Operating System CPU Storage IO
Sql Server Licensing Vmware
Hardware and drivers must be in VMware’s HCL Outdated drivers, firmware and BIOS revs adversely affect Virtualization. Always disable unused physical hardware devices Leave BIOS memory cleanup speed at default. (or equivalent) Disable CPU C-States / C1E Suspend Enable All CPU Cores – Prevent hardware from disabling cores. -x / AMD-V Memory Management Unit (MMU) Virtualization Intel Extended Page Tables (EPT) / AMD Rapid Virtualization Index (RVI) I/O MMU Virtualization Intel VT-d / AMD-Vi/IOMMU
What Does Vmware Changing Its Licensing Model Mean For Enterprise It?
Rapid-fire Fact Dump It’s a storage space, crap ALWAYS has a queue on a single lane highway and a 4 lane highway. For all data request volumes, PVSCSI is better. Ask your storage vendor about their multi-way policy. More is not better. Know the NUMA limit of the hardware. Use it to set your size Beware of memory tax, beware of CPU fairness There’s no such thing as (VM’s home node). Don’t blame the vNIC VMXNet3 is NOT the problem Outdated VMware tools may be the problem. – for example. Consider disabling RSS outage aggregation Use your tools Virtualization doesn’t change SQL administrative tasks – SQL DMV? ESXTop – Specific to ESXi Visualesxtop – Esxplot –
VSCSI Adapter Application VMKernel FC/iSCSI/NAS Virtual Adapter Queue Depth Adapter Type Number of Virtual Disks VMKernel Resolution (Disk.SchedNumReqOutstanding) Per-Path Queue Depth Adapter Queue Depth Storage Network (Link Speed, Zoning, Subnet allocation) Number of disks (HSpin) Target queues LUN queue depth array SP
Billing Customers Queue Entry Entry Exit Server Queue Time Service Time Response Time Usage = busy time on server / elapsed time
VSCSI adapter. Note the maximum queue depth values for each device/adapter (KB 1267) LSI Logic SAS = 32 PVSCSI = 64 Just increasing the queue depth NOT even for PVSCSI – At least one for data, TempDB and logs use multiple PVSCSI adapters. volumes No native Windows drivers – Always keep VMware tools up to date Requires registry key to use Windows preferences Key: HKLMSYSTEMCurrentControlSetservicespvscsiParametersDevice Value: DriverParameter | Value data: “RequestRingPages=32, MaxQueueDepth=254” Smaller or larger datastores? Datastores also have queue depth. Always remember that LUN queue depth is determined by IP storage? – Use Jumbo frames if supported by physical network devices.
Buy Sql Server 2019 License Key
VMKernel Admittance VMKernel admission policy (KB 1268) affecting shared datastore Use dedicated datastores for DB and log volumes VMKernel admission dynamically changes when SIOC is enabled Reduce IOs for low-level VMs can be used to manage) Physical HBAs Follow vendor recommendations as much as possible LUN ( Follow vendor recommendation to reduce HBA performance. If host is connected to multiple storage arrays, settings are global. Consult vendor for correct multipath policy for us Yes , array manufacturers know.
Generally similar performance vSphere 5.5 and later support VMDK files up to 62 TB Disk size is no longer a limitation for VMFS VMFS RDM. Better storage consolidation – multiple virtual disks/virtual machines per VMFS LUN. But still a LUN can assign a single virtual machine Implements a 1:1 mapping between a virtual machine and a LUN Aggregate virtual machines in a LUN – vSphere LUN limit of 255 is more likely vSphere LUN limit of 255 is likely to be reached. Virtual machines on a LUN < IOPS rating of the LUN Does not affect the IOPS impact of other virtual machines. When to Use Raw Device Mapping (RDM) Required for a shared disk replacement cluster. Required by storage vendor for SAN management tools such as backups and snapshots. Otherwise use VMFS.
Features: shared DataStore/LUN 1 database OS; 4 equal size data files on 4 LUNs 1 TempDB; 4 LUNs Data, TempDB and Log files of equal size 4 (1/vCPU) tempdb files spread across 3 PVSCSI adapters Data and TempDB files can be RDMed with PVSCSI adapters Virtual disks Advantages: Optimal performance; Each Data, TempDB, and Log file has VMDK/Datastore/LUN I/O spread evenly across the PVSCSI adapters. Log traffic is incompatible with random data/TempDB traffic. Disadvantages: Windows can run out of drive letters quickly! More complex storage management
Features: shared DataStore/LUN 1 database OS; 8 equal size data files 4 LUNs 1 TempDB; 4 files (1/vCPU) are evenly distributed and mixed with data files to avoid “hot spots”. Data, TempDB and Log files spread across 3 PVSCSI adapters Virtual disks can be RDM. Advantages: I/O is spread more evenly/TempDB uses fewer disk letters. avoid hotspots Log traffic is inconsistent with random Data/TempDB traffic
Vmware Horizon Details
8 vCPU VMs with less than 45 GB per VM ESX Scheduler. If the VM size is larger than 45 GB or 8 CPUs, NUMA intermediate and subsequent migration occurs and can cause a 30% memory throughput drop. Each NUMA node has 94/2 45 GB (less 4 GB for the hypervisor) memory and 96 GB of RAM on the server.
Number of Sockets on vSphere Host vSphere Overhead Number of VMs on vSphere Host vSphere RAM Overhead 1% RAM Overhead Physical RAM on vSphere Host Physical RAM on vSphere Host
Why VMware Recommends Enabling NUMA Windows Microsoft SQL Server NUMA-Aware vSphere NUMA Benefits Use It People Enable Host-Level NUMA Disable ‘leave node spacing’ in BIOS – HP consult hardware vendor for CUSTOM virtual configuration on systems NUMA is loved by ALL MS SQL Servers worldwide. Automatically enabled in vSphere for any VM with 8 vCPUs. Want to use it on smaller VMs? Set “numa.vcpu.min” to # vCPU in VM Virtual NUMA and CPU Hot-plug. Maybe later
21 NUMA Best Practices Prevent remote NUMA access Size of vCPUs <= number of cores per NUMA node (processor socket) Align VMs with physical NUMA boundaries if possible Align NUMA boundaries for large VMs use more or even denominator. Hyperthreading Initial conservative sizing: set vCPUs equal to # physical cores HT benefit is about 30-50%, < for CPU-intensive batch jobs (based on OLTP workload tests) distribute vCPUs by number of sockets "Socket leave "core per core" at default value "1" to monitor NUMA performance in vSphere Coreinfo.exe to view NUMA topology on ESXTOP Windows Guest Between hosts with the same NUMA architecture to avoid performance degradation if vMotioning move (until reloaded)
Successfully Virtualizing Sql Server On Vsphere
1 vCPU OFF per core with Hyperthreading. 1 vCPU per core for SQL Server 10%-25% increase in processing power with hyperthreading ENABLED Uniform Licensing View HT does not change core licensing requirements “numa.vcpu.preferHT” to true. Force 24-way VM scheduling on NUMA node
Extends NUMA awareness for guest OS Multi-core enabled via UI 8+ vCPU multi-core is enabled by default for VMs. Existing VMs will not be affected by the update For smaller VMs, enable by setting numa.vcpu.min=4 CPU Hot-Add For large virtual machines, confirm that the feature is enabled for best performance
The VM itself is important – optimize in the guest Windows CPU core stall = BAD™ Set power to ‘High Performance’ to avoid core stall Windows host side scaling settings Impact CPU usage NIC and Windows kernel must be enabled at the level “use netsh int tcp show global” Follow the vendor’s recommendation to check the application-level configuration Virtualization does not change the view
2012//64-bit default ON and higher version requires “Lock pages in memory” user right for SQL Server service account (sqlservr.exe) – Implication Monitor Mitigation Slow startup due to memory pre-allocation ERRORLOG message Memory reservation Impact to RTO for FCI and OK for VMware YES AAG can help because during outage SQL will not allocate less than “max server memory” or even fail to start due to memory fragmentation ERRORLOG or sys.dm_os_process_memory Dedicate server to SQL earlier Startup SQL use possible return to default page than others Sponsored by SQLArg3 and value set “-T834” -Itemproperty -Path “HKLM:SOFTWAREMicrosoftMicrosoft SQL ServerMSSQL13.MSSQLSERVERMSSQLServerParameters” -name SQLArgT3 “3” – val Self-explanatory slide.
Microsoft Sql Server Cal Guide — Trusted Tech Team
27 Memory reserves guarantee memory for the VM – even if there is a conflict, the VM is only allowed to power on when CPU and memory are available (hard permission) If allocated RAM = reserved RAM, you avoid swapping Don’t limit the mission . -Critical SQL VMs If you use Resource Pools, put down-level VMs in Resource Pools SQL supports “Memory Hotadd” Don’t use it on ESXi versions below 6.0 if maximum memory is set for SQL instances lsa, you need to run sp_configure
Sql server 2017 licensing, sql server licensing costs, sql server 2019 licensing, microsoft sql server licensing, sql server licensing 2014, aws sql server licensing, sql server standard licensing, sql server licensing, sql server core licensing, windows sql server licensing, sql server 2016 licensing, sql server licensing guide