The Three Layers of Cloud Infrastructure
Cloud hosting is not a single technology — it is three distinct infrastructure layers working together: compute (the processing power), storage (where data lives), and network (how everything connects). Each layer has its own specifications, redundancy requirements, and impact on the performance you experience. Understanding each layer helps you evaluate hosting plans more accurately.
Layer 1 — Compute: Virtual Machines and CPU/RAM
The compute layer is the virtualisation infrastructure — physical servers running a hypervisor that creates isolated virtual machines. Each VM receives allocated CPU cores and RAM. For accounting software hosting, compute specifications affect: how many concurrent sessions the server supports (RAM), how fast intensive operations (report generation, batch reconciliation) complete (CPU), and whether performance is consistent or variable under load (dedicated vs shared resources).
M A Global Network's compute layer uses dedicated vCores and dedicated RAM — resources reserved exclusively for your VM, not drawn from a shared pool.
Layer 2 — Storage: NVMe SSD and RAID Arrays
The storage layer holds all data — Tally company files, Busy databases, application installations, OS files. Storage specifications affect: how quickly database read/write operations complete (IOPS — where NVMe dramatically outperforms SATA), whether a single drive failure causes data loss (RAID redundancy), and how much data the server can hold (capacity).
Data centres use enterprise NVMe SSD in RAID configurations. A single drive failure in a RAID-10 array does not cause data loss or service interruption — the array continues on remaining drives while the failed drive is replaced. This hardware-level protection is in addition to the daily off-site backup.
Layer 3 — Network: Data Centre Connectivity
The network layer includes: internal data centre network (25–100 Gbps between servers and storage), internet uplinks (1–10 Gbps with multiple ISP redundancy), and the managed firewall (controls which inbound connections are allowed based on source IP — the IP-whitelisting that protects your server).
For Indian users: a server in an Indian data centre has 10–30 ms round-trip latency. Servers in Singapore add 60–80 ms. All three layers operate on the same physical infrastructure in M A Global Network's Indian data centre.
How the Layers Work Together for Tally/Busy
When a user enters a voucher in Tally on a cloud server:
- Network layer: Encrypted RDP connection transmits the keyboard input from user's device to server (10–30 ms for Indian users)
- Compute layer: Tally process on the VM (using dedicated vCores) processes the input and requests a database write
- Storage layer: NVMe SSD (300,000+ IOPS) completes the database write in microseconds
- Compute layer: Tally updates the screen display
- Network layer: Updated screen image transmitted back to user's device
Total round-trip: under 100ms for Indian users on Indian-hosted servers. This is why Tally on a correctly configured cloud server feels effectively instantaneous.
Frequently Asked Questions
For accounting software hosting with managed service: yes. A single provider managing compute, storage, and network ensures consistent configuration, unified security management (firewall at network layer protecting the compute layer's services), and single-point accountability when something goes wrong. Multi-provider infrastructure splits responsibility in ways that complicate incident resolution.
All Three Layers — Managed as One
NVMe storage · Dedicated compute · Redundant network. ₹700/user/month + 18% GST. Yearly plan.