Overview

Cloud Infrastructure
Our cloud infrastructure is designed to offer scalable, efficient, and reliable computing resources. Leveraging the power of cloud computing, we ensure that our services are always available, responsive, and capable of handling evolving business needs.

Choice of Azure Cloud
Azure Cloud has been our choice for cloud services owing to its robust scalability, exceptional reliability, comprehensive security measures, and diverse range of services. Azure's global presence and continuous innovation align perfectly with our goals for digital transformation.

Key Azure Services and Features
Our architecture incorporates several Azure services:

• Azure Virtual Machines for flexible, scalable computing power.
• Azure Blob Storage for high-performance, secure data storage.
• Azure SQL Database for reliable, managed database services.
• Azure Kubernetes Service (AKS) for efficient container orchestration.

Security and Compliance
Security is paramount. We implement rigorous network security protocols, data encryption standards, and identity and access management policies. Compliance with industry standards is strictly adhered to, ensuring data protection and privacy.

Scalability and Performance Optimization
Our Azure infrastructure is built for scalability, enabling us to efficiently manage workload fluctuations. Performance optimization is a continuous process, ensuring high availability and minimal latency in our services.

---

Shared & Dedicated Cluster

HSP provides two distinct options for our API and HSP clients, catering to different business needs and scales:

Shared Cluster:

What It Is:
The Shared Cluster is a communal environment used by our clients for various API activities such as searches, bookings, and other related functions, as well as for Back-Office operations.
Who It's For:
Ideal for small to medium-sized businesses.
How It Works:
In this shared setting, clients operate alongside others. However, it's important to note that excessive use, spike or stress testing by any client in the Shared Cluster could lead to slower response times for everyone using that cluster.

👍

NOTE:

Clients with high activity within the "Shared Cluster" will be subject to throttling measures to ensure system performance is maintained and to prevent any adverse impact on other clients

Private/Dedicated Cluster:

What It Is:
The Private/Dedicated Cluster offers an exclusive environment tailored for each client.
Who It's For:
Suited for larger businesses seeking control and independence from the shared environment.
How It Works:
Operations in this cluster are isolated, ensuring your activities are not impacted by other clients. This exclusive 'space' guarantees that your processes run smoothly and efficiently.

👍

NOTE:

If you expect to have a high number of searches and would like to have your own dedicated cluster, it's doable, take into consideration it comes with costs .

Estimating Your Azure Costs

The cost analysis for your Dedicated Cluster on Azure is influenced by a variety of factors including client trends and the resources you use. Key components contributing to the cost include Virtual Machines (VMs), Storage solutions, Bandwidth usage, Virtual Networking, Load Balancers, among others.

👍

NOTE:

For an estimation of your Azure costs, please use - Azure Pricing Calculator.

For convenience - we have listed the services you're utilizing along with their respective tiers below.

---

Case study: Cost Estimation 1

Client Type:

• B2C Client.
• Running multiple campaigns.
• Client that using API users.
• Clients average Searches per day - 5,773
• Monthly average - $1,542.

---

Case study: Cost Estimation 2

Client Type:

• Client that using multiple Agents + API users.
• Client is using cache.
• Clients average Searches per day - 562,896
• Monthly average - $3,506.

---

Case study: Cost Estimation 3

Client type:

• Clients using multiple distribution channels.
• Clients using multiple Agents + API users.
• High count of searches per day.
• Clients average Searches per day - 2,004,752
• Monthly average - $5,435.

---

Optimization

As previously mentioned, the costs associated with your Dedicated Cluster on Azure are affected by several factors.
This section aims to provide guidance on optimizing these costs and your Azure usage through various methods. These include managing session and data storage, optimizing search processes, limitations, scaling, and more.

Stateless Search

Each interaction with our API, including actions like Search, GetPackages, cxl-policy, book, and others, along with transactions and bookings, are tracked and logged in our Glogs - a dedicated session viewer tool.
This data, stored on client storage, contributes to Azure costs.

For users engaged in frequent searches or those who require caching or meta-searches, we've designed the 'Stateless' Search method.
This request enables you to execute searches without saving session data and associated dumps.
By doing so, it offers a way to minimize the storage requirements for your data.

More information - Click here.

Storage

We track and log every interaction with our API, including searches, get packages, CXL policies, bookings, and more.

However, storing all this information can grow expensive. To keep costs down and optimize your Azure usage, HSP offers "Reduce Dumping".

What does "Reduce Dumping" do? It shrinks the size of your session logs, minimizing data storage requirements and lowering your Azure costs. In simple terms, you get the same great insights with less data and expense.

In order to enable the "Reduce Dumping" feature - please contact our support.

Throttling

Managing multiple distribution channels, including different agents and API users, can be difficult, particularly when tracking their usage.

To address this challenge, Throttling Settings have been implemented. These settings are crucial for preventing system abuse and controlling the usage of specific services and request types. This is especially important in scenarios where API clients might improperly implement these services.

For more information about Throttling - Click here.

Scale settings

To optimize costs efficiency and functionality in your Azure cluster, several key strategies can be addressed:

Auto-Scaling: Implement auto-scaling within the Azure cluster to dynamically adjust resource allocation in response to demand, thereby lowering costs during periods of reduced usage.

Resource Optimization: Conduct regular assessments and adjustments of the cluster's resource sizes, like VMs, ensuring they match actual usage and preventing excess expenditure on resources that are not fully utilized.

Performance and Cost Monitoring: Employ Azure Cost Management tools for ongoing monitoring of both cluster performance and expenditure. This facilitates timely and informed adjustments to scale settings.

Additionally, there are various other measures that can be taken.

Our team is committed to expertly managing and optimizing your Azure cluster, providing dedicated and professional service.

VMs (Spot & Public)

Spot and Public instances in cloud computing have distinct differences:

Pricing

• Spot Instances: The pricing for Spot instances is significantly lower, often up to 90% cheaper than Public instances. This is because they utilize the cloud provider's excess capacity. However, the prices are variable and based on real-time supply and demand, which means they can fluctuate frequently.
• Public Instances (On-Demand): These instances have fixed, predictable pricing. They are more expensive than Spot instances, but you pay a premium for the reliability and the assurance that the instance won't be terminated unexpectedly due to demand fluctuations.
  Performance:

Performance

• Spot instances: are generally equivalent to Public instances, as they use the same underlying hardware. The key difference lies in their availability and reliability. Spot instances can be terminated by the provider with very little notice if there's an increase in demand for the capacity, which makes them less reliable for critical, uninterrupted workloads.
• Public Instances (On-Demand): These offer consistent performance and availability. They are not subject to termination due to demand fluctuations, making them suitable for workloads that require stability and continuous operation.
  To optimize the use of these instances, it's important to assess your workload requirements. Use Spot instances for non-essential, flexible, or batch processing tasks where interruption is acceptable, and reserve Public instances for critical, performance-sensitive workloads that require uninterrupted operation. This approach can significantly reduce costs while maintaining the necessary performance and reliability.