If you have ever compared VPS plans and noticed that one provider offers “2 vCPU” for $3/month while another charges $29/month for dedicated cores, you probably wondered what the difference is. The answer is the single most important technical detail in VPS hosting for traders, and it is the one that most providers would rather you not understand.
What a CPU Core Actually Is
A physical server in a datacenter has a processor — typically an Intel Xeon or AMD EPYC — with a fixed number of cores. A 32-core processor can execute 32 independent threads simultaneously at full speed. Each core is a real, physical piece of silicon on the chip.
When a VPS provider creates virtual machines on that server, they allocate CPU cores to each VPS. How they allocate those cores determines everything about your performance.
Shared vCPU: The Budget Approach
With shared vCPU (also called “burstable” or “standard” instances), the provider sells the same physical core to multiple customers. A single physical core might be assigned to 4, 8, or even 10 different VPS instances simultaneously.
💡 Tip: Always confirm whether your VPS provider offers dedicated CPU cores or shared vCPUs. Shared cores mean your EA’s execution speed fluctuates based on other users’ workloads — exactly the inconsistency scalpers cannot afford.
The economics are straightforward: if a physical core costs the provider $20/month to operate and they sell it to 8 customers at $3/month each, they make $24/month. Everyone profits — as long as not everyone uses their CPU at the same time.
This is called overcommitting, and the ratio matters. Common overcommit ratios in the budget VPS market:
- 4:1 — moderate, performance is usually acceptable
- 8:1 — aggressive, noticeable slowdowns during peak usage
- 10:1 or higher — severe overcommit, performance is unpredictable
When your VPS neighbors are idle — reading email, running a static website with no traffic, sitting unused — you get access to the full physical core speed. Your VPS feels fast. The problem is what happens when they are not idle.
The Noisy Neighbor Problem
Imagine you share a highway lane with 9 other cars. At 3 AM, the road is empty and you cruise at full speed. At 5 PM rush hour, all 10 cars are on the road and everyone slows to a crawl.
Shared vCPU works the same way. When multiple VPS instances on the same physical core are all demanding CPU time simultaneously, the hypervisor (the software managing the physical server) must divide that core’s processing time among them. Your “2 vCPU” VPS might only get 25% of a real core’s processing power during peak times.
You have no visibility into this. You cannot see what your neighbors are running. You cannot predict when they will be busy. You only notice that your VPS is suddenly slower.
Dedicated Cores: What You Pay For Is What You Get
With dedicated cores, the math is different. A physical core is assigned to your VPS and only your VPS. No sharing. No overcommit. No neighbors.
If you pay for 2 dedicated cores, you have 2 physical cores reserved for your workload at all times, regardless of what anyone else on the server is doing. Your performance at 3 AM is identical to your performance during NFP. For a broader comparison of hosting tiers, see VPS vs dedicated server: which is right for you.
📊 Key Stat: Scalping strategies targeting 2-3 pip moves lose 10-15% of their edge per trade on a 50ms+ connection. Sub-5ms latency makes network slippage negligible.
This costs the provider more because they sell each core only once. That cost is passed to you. But in return, you get predictable, consistent performance.
Why This Matters Specifically for Trading
General web hosting workloads are “bursty” — a website needs CPU for a fraction of a second when a page is requested, then sits idle. Shared vCPU works fine for this because the bursts rarely overlap across customers.
Trading workloads are the exact opposite. They have a specific, predictable pattern: everyone gets busy at the same time.
The NFP Scenario
Non-Farm Payrolls releases at 8:30 AM Eastern on the first Friday of the month. In the 5 seconds following the release:
- Every scalper’s EA is processing a flood of tick data
- Every news trader’s algorithm is analyzing the deviation from forecast
- Every breakout strategy is evaluating price levels
- Every risk management system is calculating new position sizes
⚠️ Warning: During major news events (NFP, FOMC, ECB), tick volume spikes 5-10x and spreads widen dramatically. Ensure your VPS has CPU headroom — if CPU hits 100% during news, orders queue instead of executing.
If you are on a shared vCPU server with other traders (which is likely — trading VPS providers attract traders), every VPS on that physical server demands maximum CPU at the exact same moment. The hypervisor divides the physical core’s time among all of them, and everyone gets degraded performance precisely when they need peak performance.
Real-World Performance Comparison
On a shared vCPU during a volatile market event:
- EA tick processing time: 2-15ms (variable, depends on neighbor load)
- Order preparation and submission: 1-5ms additional processing delay
- Total CPU contribution to execution: 3-20ms, unpredictable
On a dedicated core during the same event:
- EA tick processing time: 0.1-0.5ms (consistent)
- Order preparation and submission: 0.1ms processing delay
- Total CPU contribution to execution: 0.2-0.6ms, predictable
The difference is not just speed — it is consistency. With shared vCPU, you cannot predict your execution time. With dedicated cores, you can.
For scalpers, this matters on every single trade — see scalping and high-frequency trading on a VPS for a deeper look. For EA traders, it compounds across hundreds of trades per month. For news traders, it is the difference between catching the move and missing it entirely.
✅ Best Practice: Strip your MT4/MT5 to the bare minimum — only the charts and symbols your EA needs. Every extra element competes for CPU cycles your orders require.
How to Tell If Your VPS Has Dedicated Cores
If your provider does not explicitly state “dedicated cores” or “dedicated CPU,” assume it is shared. The terminology varies — “vCPU,” “virtual cores,” “CPU cores” without the word “dedicated” — but shared is the default in the industry because it is cheaper to provision.
You can also test it yourself:
Run a CPU benchmark multiple times over several days. Use a tool like Cinebench or even a simple CPU stress test. If your score is consistent (within 5% variance), your cores are likely dedicated. If the score varies by 20-50% depending on when you run it, you are sharing.
Test during high-volatility events. If your VPS noticeably slows down during NFP, FOMC, or ECB announcements — when other traders on the same server are also busy — you are on shared vCPU.
Check the price. A VPS with 2 dedicated cores, 4GB RAM, and NVMe storage in a financial datacenter has real costs. If a provider is offering that for $2-3/month, they are overcommitting aggressively. The infrastructure simply cannot be provisioned at that price point without sharing.
The FXVPS Approach
All FXVPS plans use dedicated CPU cores. No overcommitting. Your cores are reserved for your VPS and are not shared with other customers.
We maintain a maximum 1.5:1 CPU allocation ratio across our infrastructure. In practice, this means we keep significant headroom on every physical server. We could pack more VPS instances onto each server and charge less. We choose not to, because the moment you experience a slowdown during a volatile market is the moment you lose money — and lose trust in your infrastructure.
This is why our Core plan starts at $29/month. It buys you CPU resources that are genuinely yours. For a trader risking real capital, that is not a cost — it is the cheapest insurance available.
The Bottom Line
Shared vCPU is designed for workloads where performance fluctuations are tolerable: websites, development environments, personal projects. Trading is not one of those workloads.
When your EA needs to process a tick and submit an order, it needs CPU immediately, not “when the hypervisor gets around to scheduling your time slice.” Dedicated cores guarantee that the CPU is available the instant your software needs it.
For swing traders who hold positions for days, shared vCPU might be adequate. For scalpers, EA traders, and anyone running automated strategies during volatile sessions, dedicated cores are not a luxury — they are a requirement for consistent execution.
Check your current VPS provider. If you are on shared vCPU, benchmark it during the next NFP release. Compare your EA’s execution times to normal market hours. If you see degradation, you know what is happening — and you know the fix.