Mastering Custom Queues for Compute Engine VM Performance

Making Use of Custom Queues for Each vNIC on the Compute Engine

[{"selector":"#anim-a8e833c2-8648-45d7-a8fc-409942648543 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(34.64226964365199%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-9288a3bb-f5ff-4970-8583-25215d3418c5","keyframes":{"transform":["translate3d(137.91667%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":1000,"easing":"cubic-bezier(.2, 0, .8, 1)","fill":"both"}] [{"selector":"#anim-42840681-a468-4004-ab51-a96381ef6065","keyframes":{"transform":["rotateZ(180deg)","rotateZ(0deg)"]},"delay":0,"duration":1000,"easing":"cubic-bezier(.2, 0, .5, 1)","fill":"forwards"}]

When companies are using network virtual appliances (NVAs), traffic-intensive AI/ML workloads, or high-bandwidth apps, it is critical to make sure your virtual machines (VMs) can support the demand

[{"selector":"#anim-8d87a958-44ba-4dbd-b25f-09cab4794214 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(21.874999829338595%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-83285b91-9345-444e-ac1f-81d36be2a20d","keyframes":{"transform":["translate3d(-119.52862%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":1000,"easing":"cubic-bezier(.2, 0, .8, 1)","fill":"both"}] [{"selector":"#anim-64604433-8417-4b31-8490-85d8e9899e18","keyframes":{"transform":["rotateZ(-180deg)","rotateZ(0deg)"]},"delay":0,"duration":1000,"easing":"cubic-bezier(.2, 0, .5, 1)","fill":"forwards"}]

Due to this, you were forced to choose between lowering the number of vNICs to guarantee that each vNIC has enough queues or vertically scaling up to a larger VM Suboptimal scaling and higher expenses resulted from both strategies

[{"selector":"#anim-7f792584-cce7-4990-8160-1ba866a48166 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(33.84712828036338%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-971f0887-8acd-41ca-ac98-fe788b97f501","keyframes":{"transform":["translate3d(-118.77133%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":1000,"easing":"cubic-bezier(.2, 0, .8, 1)","fill":"both"}] [{"selector":"#anim-b15a167a-c078-47da-8d17-9f7f1cccf95b","keyframes":{"transform":["rotateZ(-180deg)","rotateZ(0deg)"]},"delay":0,"duration":1000,"easing":"cubic-bezier(.2, 0, .5, 1)","fill":"forwards"}]

Custom queues for each vNIC are now usable. You can manually assign each virtual network interface (vNIC) up to the number of network queues that the driver type can support thanks to this capability

[{"selector":"#anim-f5190a54-da01-4c82-b3e2-dfe1ff4e3c5d [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(35.937499914669296%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}]

These workloads frequently produce enormous volumes of data, which custom queues can help to ensure that your virtual machines (VMs) can handle and transport without experiencing any delays

[{"selector":"#anim-03a944bd-aa72-444e-bdc9-3d08de2cb8d2","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-a14e4b8b-1bf9-48e8-92ec-c9460b1db104","keyframes":{"transform":["translate3d(0px, -112.30361%, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}]

Gen 2 GCE virtual machines and Tier 1 high-bandwidth networking come with custom queues, thus utilizing this potent feature doesn’t cost extra

[{"selector":"#anim-0dd87989-caca-454f-9cfc-951d3c9c57b3 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(31.284090795523493%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-5077812e-8f57-4d2f-afd5-e30128739e7c","keyframes":{"transform":["translate3d(-129.72973%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":1000,"easing":"cubic-bezier(.2, 0, .8, 1)","fill":"both"}] [{"selector":"#anim-6f765fef-db57-490f-8d06-837dcb793079","keyframes":{"transform":["rotateZ(-180deg)","rotateZ(0deg)"]},"delay":0,"duration":1000,"easing":"cubic-bezier(.2, 0, .5, 1)","fill":"forwards"}]

Google were able to push an aggregate throughput of about 50% of the NVA VM’s stated throughput using the usual queue allocation, but with queue oversubscription, this nearly reached 99% of the NVA VM’s stated throughput

[{"selector":"#anim-5f6f0ec6-2306-453d-b347-384087ce2118 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(35.937499914669296%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-dbaa476a-bf11-4e1e-b247-3b6bcf88d1f8","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-176aa585-b266-404a-b4fb-3431985a76d8","keyframes":{"transform":["translate3d(0px, -111.84186%, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}]

This functionality is especially beneficial for traffic-intensive AI/ML workloads, high-bandwidth apps For more details visit govindhtech.com

[{"selector":"#anim-a3d6f905-576f-497d-992a-f300a4f2bc48 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(34.64226964365199%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}]