Table of Contents
- 1. The four slices, and the CPU rule on each
- 2. The High Volume line, which is where the port speed is stated
- 3. What "unmetered" does and does not say
- 4. Block storage at $5 a terabyte
- 5. Where BuyVM lists the servers
- 6. Refunds and cancellation
- 7. What BuyVM does not publish
- 8. Measuring the parts that are not published
- FAQ
- Sources
1. The four slices, and the CPU rule on each
BuyVM calls these dedicated KVM slices and states three things on every card: the core count with a clock, whether the CPU is shared, and the storage type ("100% SSD Storage" in the page title). The CPU line is unusually specific for this market, and it is also where the purchase decision sits.
| Plan | Memory | SSD | CPU usage | Price as printed | Per month |
|---|---|---|---|---|---|
| SLICE 512 | 512 MB | 10 GB | Fair share | $24.00 per year | $2.00 |
| SLICE 1024 | 1024 MB | 20 GB | Fair share | $3.50 per month | $3.50 |
| SLICE 2048 | 2048 MB | 40 GB | Fair share | $7.00 per month | $7.00 |
| SLICE 4096 | 4096 MB | 80 GB | Dedicated | $15.00 per month | $15.00 |
The clock reads "3.50+ GHz" on all four plans, which is a floor rather than a processor model. The sharing rule is the part that changes: the three cheapest plans say "Fair Share CPU Usage" and the 4 GB plan says "Dedicated CPU Usage", so the $7 to $15 step buys an unshared core as well as the memory. Every plan carries one IPv4 address and unmetered bandwidth, and the smallest is priced per year with no monthly figure shown.
2. The High Volume line, which is where the port speed is stated
Above the slices there is a second ladder, linear at $15 a core, with the port speed printed in the table rather than described in prose.
| Cores | Memory | SSD | Bandwidth | Monthly |
|---|---|---|---|---|
| 2 @ 3.50+ GHz | 8 GB | 160 GB | Unlim. 1000 Mbps | $30.00 |
| 3 @ 3.50+ GHz | 12 GB | 240 GB | Unlim. 1000 Mbps | $45.00 |
| 4 @ 3.50+ GHz | 16 GB | 320 GB | Unlim. 1000 Mbps | $60.00 |
| 5 @ 3.50+ GHz | 20 GB | 400 GB | Unlim. 1000 Mbps | $75.00 |
| 6 @ 3.50+ GHz | 24 GB | 480 GB | Unlim. 1000 Mbps | $90.00 |
| 7 @ 3.50+ GHz | 28 GB | 560 GB | Unlim. 1000 Mbps | $105.00 |
Every plan in that table carries the same bandwidth line, "Unlim. 1000 Mbps", so the port is the same at $30 as at $105 and the ladder scales cores, memory and disk only. At 2 cores and 8 GB for $30 the shape is 4 GB per core and 80 GB of disk per core, and it holds all the way up.
3. What "unmetered" does and does not say
The word appears on every slice card without a number beside it. On the pages read for this one that is the whole statement on the plan cards: no transfer allowance and no port speed. BuyVM's acceptable-use policy sets out prohibited uses in detail, from TOR exit nodes (welcome if you open a ticket) to mail policy, and defines resource abuse as use that "consumes sufficient system resources to negatively affect other clients or equipment". It asks anyone whose use might be too intense to open a ticket. It does not attach a transfer number to the unmetered claim.
That matters because "unmetered" and "unlimited" are not the same promise. An unmetered plan does not bill by the terabyte; it may still be limited by the port, by a fair-use rule applied at the provider's discretion, or by nothing at all. At BuyVM the only published figure in this area belongs to the other product line, 1000 Mbps on High Volume. A reader who needs a stated port figure for a slice should ask support before ordering rather than assume it matches.
4. Block storage at $5 a terabyte
The block storage slabs are the most quotable number in BuyVM's catalogue: "Order 1TB of usable storage for just $5.00 / month". BuyVM describes the platform as a "40Gbit+ InfiniBand RDMA Storage Fabric giving near local storage performance", and says a slab can "be detached, and reattached to a KVM Dedicated Server Slice located within the same facility without downtime or reboots".
The arithmetic against the slices is the reason to care. SLICE 2048 gives 40 GB of SSD for $7.00 a month, which is $0.175 a gigabyte. A slab is $0.005 a gigabyte. The slab is therefore about 35 times cheaper per gigabyte, which is why a server that needs to hold a lot of data buys a small slice and attaches storage to it rather than buying a bigger slice. What it is not is local disk: it attaches over the storage fabric, it has to sit in the same facility as the server, and it cannot be shrunk once ordered.
The pricing is linear and sold in units rather than sizes: slabs come "in chunks of 256GB for $1.25/month", so 1 TB is $5.00 and the 10 TB slab is $50.00, and a buyer who needs 3 TB orders twelve units. The limits are published on the same page. A slab can be at most 10 TB, a single KVM slice can have up to 8 slabs attached, downgrading is not possible ("will require you to order a new Block Storage Slab and transfer your data yourself"), and a slab cannot be bought on its own: it needs an accompanying slice in the same facility. At the maximum, eight 10 TB slabs put 80 TB behind one server.
BuyVM also publishes a comparison table on that page pricing other vendors' block storage per terabyte: OVH at $90.00, Vultr at $100.00 and Amazon EBS at $131.50, against its own $5.00. Those three figures are BuyVM's claims about its competitors, read from BuyVM's page rather than from the competitors' pages, and this page repeats them as such.
5. Where BuyVM lists the servers
The KVM slices FAQ lists three locations: Las Vegas, New York and Luxembourg. The block-storage FAQ lists those three and Miami. The slabs attach within a facility, so the storage location has to match the server's. What neither page states is which of the four slices is stocked in which location; provisioning pages for small hosts usually answer that at the order form rather than in the catalogue. The US datacenter guide maps the US cities to the latency each one suits.
6. Refunds and cancellation
Frantech's terms give a three-day window on two payment methods. PayPal: "PayPal transactions are eligible for refund within 3 days of the initial order payment. A ticket must be submitted to the Billing Department to request a refund." Cards, in the next clause: "Credit Card transactions are eligible for refund within 3 days of the initial order payment. A ticket must be submitted to the Billing Department to request a refund." The same document states that cryptocurrency and Alipay payments are not eligible for refunds. Account credit "may not be withdrawn or 'cashed out' in the form of a refund" and is not transferable, so money added as credit is spent with BuyVM rather than returned.
The other direction is a pro-rata refund: "If we cancel your service in its entirety without cause, we will refund to you on a pro-rata basis the amount of your payment corresponding to the portion of your service remaining". That sits beside a clause reserving the right to cancel or suspend at any time and for any reason, and a line making the customer responsible for their own backups. With the cheapest slice billed annually, those two paragraphs are what a buyer is agreeing to.
7. What BuyVM does not publish
- A processor SKU. Every plan card reads "1 Core @ 3.50+ GHz" or the High Volume equivalent. The slices page also says the nodes are "Powered by 3.5Ghz+ Intel processors" and does not name a model.
- A port speed for the slices, and no transfer figure either: the slice cards say "Unmetered Bandwidth" and stop.
- A transfer number attached to unmetered. The acceptable-use policy defines resource abuse and asks heavy users to open a ticket; it does not publish a terabyte ceiling.
- Disk IOPS or throughput for the slice SSDs, and no figure for the slab fabric beyond the "40Gbit+ InfiniBand" description.
- A stated latency for a slab against the slice's own disk, which is the part of "near local storage performance" a buyer would most want quantified.
- Which slices are in stock in which location, on the pages read for this one.
- An uptime figure or SLA on the slices page, the block storage page or the terms.
8. Measuring the parts that are not published
The processor, the disk and the achievable throughput are measurable on the server itself. A buyer who wants those figures can run lscpu for the CPU model and clock, sysbench cpu --threads=1 --time=60 run for single-thread speed, a 4K fio run for the disk, and iperf3 against a machine near the users for the network. This site publishes none of those measurements. The benchmarks hub has the commands and the reasoning, the storage page compares what other providers publish, and the BuyVM review covers how the slices compare with the rest of this site.
Two measurements are worth making here specifically. The first is the port speed, because the slices do not state one: a single iperf3 run settles it. The second is the slab's latency against the slice's own disk, because the claim is "near local storage performance" and that is measurable with fio on both volumes from the same server.