IBM Cloud Bare Metal Servers Review

IBM Cloud Bare Metal Servers were being used at all levels of our infrastructure. The uniqueness of IBM's offering bridges hardware and global networks into a nice package. IBM is unique as it has the ability to bind IP addresses that can float across any data center region to allow for fault-tolerant applications bound by a static IP regardless of geographic location. IBM's [and formerly Softlayer's] global private network allows for data centers to interact and communicate between regions without requiring VPN or other networking tools. Allowing your application to scale globally while physically limiting intra-networking on a private VLAN prevents many other challenges of securing private data. [I feel] in the telecom space, port 5060 SIP is a widely known and 'attackable' port. While you want to allow your customers access wherever they may be, keeping 'bad actors' away can be a challenge at scale. Our platform was physically built for the IBM Cloud to scale within its private network, keeping the intra-workings between all components private while only allowing known customers access on an independent transport bound to an IP that could self-heal and migrate into any global region on the IBM Cloud autonomously.

The machines performed quite well while working until they didn't [in my experience].

While base servers start at very high performance, offloading application workloads into IBM Cloud would be beneficial. Keeping in mind the cost of [computing] is greater than other vendors. The benefit of offloading workload in the IBM Cloud should be calculated based on its dependency on other network of specialty driven capability of IBM Cloud.

Compute performance does not have benefit over other compute offerings based on price alone.

[In my experience,] this is not the case [and] provisioning took at least 8 hours the first time. [In my experience,] over 16 hours when [the] system failed and had to [rebuilt.] There may be new provisions in place due to our event that requires firmware updating on all components before delivery which now can take machines to be provisioned in far longer times. [I feel] there is no 2-4 hour delivery time.

One new server we deployed took, [in our experience] almost 48 hours to deliver, however, we experienced boot failures and were only able to get into the operating system while selecting the boot disk device manually at start up. This means in the event of a failure or reboot, the system would not recover to OS. IBM helpdesk made bios adjustments beyond what is allowed within the IMPI management, however, the system never booted properly. Had to cancel months later.

Implementation on some equipment was smooth. [In my experience,] deployment time always exceeded advertised time. In one instance, our server deployment took almost 48 hours when (2-4 hour deployment time was advertised). However [it] never booted correctly and after several attempts to have IBM support repair, we had to cancel those servers.

I believe deployment times have been updated more accurately now to reflect 24-48 hour deployment times.

We did not purchase additional support initially, however on future servers, we did add premium support and insurance.

[In my experience,] No.