All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Oracle's hard drives make me feel shocked, does yours do the same?
Tue Dec 3 08:21:04 AM UTC 2024
ARM compatibility is considered experimental
Basic System Information:
Uptime : 25 days, 9 hours, 46 minutes
Processor : Neoverse-N1
BIOS virt-7.2 CPU @ 2.0GHz
CPU cores : 4 @ ??? MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ❌ Disabled
RAM : 23.4 GiB
Swap : 0.0 KiB
Disk : 195.8 GiB
Distro : Debian GNU/Linux 12 (bookworm)
Kernel : 6.1.0-26-arm64
VM Type : KVM
IPv4/IPv6 : ✔ Online / ✔ Online
IPv6 Network Information:
ISP : Oracle Corporation
ASN : AS31898 Oracle Corporation
Host : Oracle Corporation
Location : Singapore, South West (05)
Country : Singapore
fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/sda2):
Block Size | 4k (IOPS) | 64k (IOPS) |
---|---|---|
Read | 47.79 MB/s (11.9k) | 109.01 MB/s (1.7k) |
Write | 47.76 MB/s (11.9k) | 112.25 MB/s (1.7k) |
Total | 95.55 MB/s (23.8k) | 221.26 MB/s (3.4k) |
Block Size | 512k (IOPS) | 1m (IOPS) |
------ | --- ---- | ---- ---- |
Read | 86.70 MB/s (169) | 84.38 MB/s (82) |
Write | 94.11 MB/s (183) | 94.15 MB/s (91) |
Total | 180.81 MB/s (352) | 178.54 MB/s (173) |
iperf3 Network Speed Tests (IPv4):
Provider | Location (Link) | Send Speed | Recv Speed | Ping |
---|---|---|---|---|
Clouvider | London, UK (10G) | 235 Mbits/sec | 514 Mbits/sec | 315 ms |
Eranium | Amsterdam, NL (100G) | busy | busy | 319 ms |
Uztelecom | Tashkent, UZ (10G) | 616 Mbits/sec | busy | 249 ms |
Leaseweb | Singapore, SG (10G) | 4.01 Gbits/sec | busy | 1.52 ms |
Clouvider | Los Angeles, CA, US (10G) | 460 Mbits/sec | 875 Mbits/sec | 194 ms |
Leaseweb | NYC, NY, US (10G) | 1.03 Gbits/sec | 758 Mbits/sec | -- |
Performing IPv4 iperf3 recv test from Edgoo (Attempt #1 of 3)...
Edgoo | Sao Paulo, BR (1G) | busy | 537 Mbits/sec | 351 ms
Comments
Its a free service, so what?
doubt the hypervisor knows about his account tier status
Totally shattered my impression of Oracle as a company.
It's a charity.
Most people's actual needs won't feel or notice a difference. Unless your use case specifically requires fast disks, most won't notice or care.
Benchmarking and comparing numbers is a LET thing. Average consumers don't do this.
Disk performance is similar to my laptop and it works just fine. I'd host a low traffic site or entry level project on a free oracle VM and be happy.
I'm going to put it on my personal blog to write articles or use it for testing purposes, I really can't seem to do much else with it.
Seems your bar is too high. If the vps is up, we already consider that a win here in LET, if the ssd is still accessible with all the 100/200 lousy neighbours, well we consider that double win. Who cares about read/ write speed.....
I think there is an option to have faster disks should you need it. I am pretty happy with OCI for my needs.
volumen-edit-io-max
Totally normal for big cloud providers - meh throughout and iops, but always consistent and with a tight latency range.
If you compare it to ceph ssd providers, which is the only fair comparison, it absolutely blows them out of the water in real world perf and latency
I'll look into it.Ha ha.
It's supposed to be limited to 10TB of network traffic and can be used to run some simple automated programs.
You seem to have a point.
I want performance i pay for dedicated servers and create clusters instead of relying on "cloud vm".
IOPS can be configured I believe. Maybe paid tiers have higher IOPS limit ?
I'm going to use a dedicated server, I just saw this server from Oracle that I applied for earlier and wanted to try it out to see how it really works because I've used Oracle's services at work before and felt very noble, so this shocked me even though it's free. 。。。。。
Also if you haven’t already, you can get the best free performance by using a single 200gb volume with maxed out VPUs
Might be something Oracle limitations. But then Oracle did introduced vpu for max performance and you can "purchase" additional vpu for the faster IO.