I am reviewing our UCS configuration and am reminded that we are using all of our port licenses on the interconnects, so when it's time to expand with another storage array, it looks like we get to pay thousands of dollars to use additional ports on the equipment we already paid for. Why is this a thing? How can a switch be sold with the expectation that you can only use a fraction of the ports and then have to pay a lot of money to use more ports?
Is anyone else sharing the same irritation? I noticed FS.com has some pretty inexpensive high-throughput switches that might work well if we decide to move toward hyper-converged infrastructure and begin transitioning away from traditional converged infrastructure. Does anyone have any thoughts/experiences they could share regarding the hyper-converged strategy?
Not that I'm itching to get away from vSphere, but Microsoft's new S2D with Hyper-V looks interesting - like they have finally come full circle and fixed the issues of the pre-virtualization days so that we are back to stacking pizza boxes in the rack. I saw a demo a while back where they were pushing 1 million IOPS using 4 Dell R730's with NVMe and RDMA NICs while running VMFleet. Those numbers sound absurd, but if true, wow! Has anyone lab'd that up? Any shops running Microsoft's baked-in hyperconvergence?
No comments:
Post a Comment