All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Vultr.com 100% SLA :: Their money where their mouth is?
Anybody seen the recently new Vultr.com 100% SLA?
- 100% Host Node Uptime
- 100% Network Uptime
- 100% Power Availability
Those are the 3 main points that make up their SLA commitment. I'm not sure what the "100% Power Availability" is about - this is a virtualised environment, there isn't going to be a circumstance where the power is out but the host node is up.
100% Uptime Guarantee - Vultr offers a 100% uptime guarantee via this Service Level Agreement based on network, power, and instance availability. Collectively, these guarantees may be referred to as the “SLA.” This SLA is provided as a supplement to the Hosting Terms and Conditions You agreed to in becoming a Vultr customer, which is hereby incorporated by reference as an indispensable part of this SLA.
This uptime guarantee is applicable on a per-service-item basis and is not applied to Your entire invoice. For example, if You have fifty (50) servers with Us, and one (1) of those servers experiences downtime, any credit due to You under this uptime guarantee would be proportional to that one (1) server’s downtime and not Your entire account with Us. This uptime guarantee does not apply to the accessibility of Vultr’s web property, DNS servers, API, or control panel.
Pretty standard stuff there, to be expected. Though note that the SLA doesn't cover their hosted DNS service, arguably more important than instance availability? As it stands, if your DNS is important to you, you should host it yourself on an instance.
Packet Loss and Latency - Vultr does not proactively monitor the packet loss or transmission latency of specific customers. Vultr does, however, proactively monitor the aggregate packet loss and transmission latency within its LAN and WAN. In the event that Vultr discovers (either from its own efforts or after being notified by You) that You are experiencing packet loss in excess of one percent (1%) (“Excess Packet Loss”) between your instance and one hop from Vultr’s border router(s) (first hop of egress providers router) and You notify Vultr via a support ticket (or Vultr has notified You of an event), Vultr will take all actions necessary to determine the source of the Excess Packet Loss/Latency.
Limitations - The uptime guarantee ONLY applies to network and instance availability during normal operation. The uptime guarantee does NOT apply to server-side software uptime. Any outage due to server software, operating systems, improper configurations, denial of service attack against your instance, instance suspension, instance paused/halted for any reason, or any other non-network or non-“host node” outage, for any reason and whether or not such outage is caused by Us due to upgrading, troubleshooting or performing any other tasks, is not subject to this uptime guarantee. The guarantee does NOT apply when a maintenance window is scheduled with a minimum of twenty-four (24) hour notice as long as the outage/packetloss does not exceed ten (10) minutes.
This I like a lot - but does it extend to IPv6 connectivity? I recently observed intermittent packet loss (~90%) over a 3-day period with IPv6 connectivity on my instance in London. Support were clueless and unwilling to acknowledge any problem, despite the issue spanning London, Paris and Frankfurt data centres. The issue started on a Saturday morning and was eventually resolved on a Monday evening, seemingly nobody is able to resolve major incidents during the weekend. My last ticket response was ignored too. All of this in mind, considering that SLA credits remain the sole discretion of Vultr, does this SLA hold much clout?
Credit payments are certainly generous.
Comments
Pardon my ignorance but, except the not-always-too-reliable 3rd party monitoring sites, how do you calculate the SLA?
And how do they?
Proof. One thing you don't have and vultr doesn't have to give you. They win whether its online 0% or 100%
Reminds me a bit of the GVH "Total-100%-Everything-SLA".
This is where it will all go wrong.
Last month node crashed and my server was offline for two days.
I really didn't ask to give me some credits but instead informing client what happened and what they are doing to fix issue are important things for me instead of default replies node unavailable our staff checking it now.
Matter of writing a script that automatically opens a support ticket once an outage is detected, and that you later manually append more information to.
I suspect though that the process is designed in part at least on the basis many will not open the ticket to claim the credit.
I am sure, they have a API.
So ping it, if no ping, create ticket. Profit.