System Requirements for Sensors Analytics Standalone Version?

• The following configuration requirements are for reference only. For specific configurations, please consult Sense Analytics technical pre-sales/support engineers to obtain accurate assessments based on customer's actual application scenarios.
• The standalone version has the following requirements for machine configuration:
  • CPU: 8 cores or more (excluding hyper-threading), supporting SSSE3, SSE4, SSE4.2, POPCNT, AVX, and AVX2 instruction sets;
    Memory: 64G or more;
  • Operating System: It is recommended to use CentOS 7.9.
  • Kernel Version: It is recommended to use the 3.10.0-1160 system version.
  • File System: xfs
  • Hard Drive: The available space of the machine's system partition should be larger than 40G, and the capacity of the data disk should be larger than 1TB, with the following requirements
    • If it is a physical disk, configure RAID1 or higher level to ensure data security;
    • If it is a cloud disk, it needs to be SSD or other disks with a throughput of more than 150MB/s;
  • Load Balancing: Provide a dedicated front-end load balancing system.
  • It does not support mixed deployment with other services;
  • Installation requires root permission.

System Requirements for Sense Analytics Cluster Version?

• The following configuration requirements are for reference only. For specific configurations, please consult Sense Analytics technical pre-sales/support engineers to obtain accurate assessments based on customer's actual application scenarios.
• The cluster version is divided into Mini Cluster (3 nodes) and Standard Cluster (3+N nodes). The standard cluster has better performance and scalability compared to the Mini Cluster.
• The cluster version has the following requirements for machine configuration:
  • At least 3 nodes (Mini Cluster);
  • CPU: 16 cores or more (excluding hyper-threading, can be reduced to 8 cores for testing), supporting SSSE3, SSE4, SSE4.2, POPCNT, AVX, and AVX2 instruction sets;
  • Memory: 64G or higher;
  • Hard Disk: The available space of each machine's system partition is greater than 40G, and the data disk requirements are as follows
    • If it is a physical disk:
      • 2 system disks/metadata disks: Adopt RAID1 or higher level, a single disk capacity of more than 1TB, and sequentially read and write at a speed of more than 150 MB/s, it is recommended to use SSD disks;
      • 4 or more data disks: No RAID or LVM, single disk capacity of more than 2TB, and total throughput of more than 600MB/s;
    • If it is a cloud disk:
      • 1 metadata disk: SSD cloud disk, or other disks with IOPS above 5000, capacity of more than 1TB;
      • Multiple data disks: SSD cloud disk, or other disks with a total throughput of more than 600MB/s, single disk capacity of more than 1TB;
  • Operating system: It is recommended to use CentOS 7.9 version
  • Kernel version: It is recommended to use version 3.10.0-1160
  • File system: xfs
  • Network: Gigabit (>=5 nodes, 10 Gigabit is required), all machines are on the same switch and can ssh to each other. If it is a cloud server, network optimization should be enabled as much as possible.
  • Load Balancing: A separate front-end load balancing system needs to be provided.
  • It does not support mixed deployment with other services;
  • Installation requires root privileges.

How much disk space is needed to store user behavior data in Sensors Analytics?

• Based on experience, 100 million pieces of user behavior data are about 40GB before compression and about 10GB after compression. In the cluster edition, which uses three replicas by default, 100 million data records are about 120GB before compression and about 30GB after compression.

How much bandwidth is needed to collect user behavior data with Sensors Analytics?

• The server bandwidth is mainly affected by the peak number of online users. The data collection service requires much more downstream bandwidth than upstream bandwidth of the server. For App data collection, it compresses and reports data every 15 seconds by default. Based on experience, each user generates an event every 5 seconds on average. Therefore, a server with 10Mbps downstream bandwidth and 1Mbps upstream bandwidth can support about 50,000 peak online users. For JavaScript collection, due to real-time reporting and no compression of data, a server with 10Mbps downstream bandwidth and 1Mbps upstream bandwidth can support about 20,000 peak online users.

Can you provide technical services related to data?

• We also provide the following technical services for a fee in addition to our products: helping clients break down data needs and determine core metrics; helping clients organize data sources and generate standardized log data based on data needs; helping clients generate dashboards to monitor various core metrics based on our products; and assisting with specific data analysis tasks according to client requirements.

Why can't your demo import data?

• The demo environment is mainly used to showcase the querying, visualization, and analysis capabilities of Sensethink Analytics. The imported data is automatically generated by our backend for demonstration purposes. If you want to try importing your own data for querying and analysis, you can contact us to get the corresponding trial environment.

How does Sensors Analytics ensure the reliability of customer data?

• Both the cluster edition and standalone edition of Sensethink Analytics use various solutions to ensure data reliability:
  • Cluster edition:
    • The cluster edition of Sensors Analytics uses a distributed storage solution with triple redundant storage for each piece of data, ensuring high data reliability.
  • Standalone edition:
    • For customers using cloud servers, we recommend using cloud disks provided by IaaS (such as UCloud's cloud SSD disks) to ensure high disk reliability.
    • For customers using physical machines, we also recommend using underlying storage solutions of at least RAID10 to ensure high disk reliability.
    • Sensors Analytics also suggests that customers package and back up short-term data at the application layer.
  • With these solutions in place, we believe that we provide high data reliability for customer data.

How does Sensors Analytics ensure the security of customer data?

• We will answer this question based on two deployment methods: private deployment and cloud edition.
  • Private Deployment:
    • In terms of data storage, this part of the data is stored on the customer's own machine, so there are no security concerns.
    • In terms of query interfaces, we can provide query services only to the internal network, and there is also login permission authentication.
    • In terms of data transmission, if backend data transmission is chosen, it is transmitted within the internal network, so there are no security concerns. If frontend tracking is used, we support  reporting data via HTTPS and then forwarding it within the internal network.
  • Cloud Edition:
    • In terms of data storage, we offer a dedicated deployment solution with complete physical isolation of data between different clients.
    • From the query interface, different clients have different permission account systems and different query tokens.
    • From data transmission, different clients use different data reporting tokens, and we also support   HTTPS data reporting.

How should the server configuration be selected for the private deployment version of Senssun Analytica?

• Sensors Analytics supports running on servers with different configurations and scales, and the factors that affect server configuration selection are actually very complex, including but not limited to:

  • The amount of data imported per day, for example, about 10 million user behaviors need to be imported every day, and the growth rate is about 20% per month.
  • The amount of accumulated historical data, for example, there are already 3 years and a cumulative total of 10 billion historical data, and it is hoped to be imported within one week.
  • Query analysis scenarios, for example, most query analysis uses data from the past month and expects results within 10 seconds; a few query analyses use several months or longer data and expect results within one minute; sampling queries can be used in some scenarios.
  • Platform usage scenarios, such as the need to integrate with CRM, marketing push systems, or offline data mining based on Sensors analytics data.

  Sensors Analytics consultants will provide detailed evaluations and give the most suitable server configuration recommendations based on the above information. If the user is unable to provide sufficient information temporarily, we will recommend based on typical usage scenarios. Since Sensedata Analytics can easily support smooth scalability from millions to hundreds of billions of data size, we usually suggest starting with a lower configuration and adjusting as needed based on the actual usage of the system to maximize cost savings for the user.

What factors affect the query performance of Sensors Analytics? How to improve query performance?

• The query performance of Sensors Analytics is influenced by the following factors:

  • The smaller the amount of data scanned in a single query, the faster the query. It should be noted that the total amount of data stored in the system does not affect the query speed of Sensors Analytics. The query speed only depends on the amount of data involved in the specific query. Generally, a funnel analysis spanning 40 days with 6 events will be much slower than an event analysis spanning only 7 days with 1 event.
  • The better the disk reading performance of the machine, the faster the query. Without considering the application layer cache, each query of Senssun Analytics needs to scan data from the disk. Therefore, the scanning performance of the disk directly affects the query speed.
  • The better the CPU performance of the machine, the faster the query. After reading data from the disk, a complete query also requires some aggregate calculations. Therefore, a better CPU performance will lead to faster queries.
  • The more machines in the cluster, the faster the query. In other cases where the hardware is the same, increasing the number of machines is equivalent to directly improving the disk reading and CPU computing performance, which will make the query faster.

  If you need to improve the query performance of Senssun Analytics, you can start with the following solutions:

• Use sampling queries to reduce the amount of data scanned in a single query. Sampling queries can be used to quickly discover and verify issues, and then close the sampling to obtain the most accurate data when needed.
• Improve machine performance. This includes replacing better disks, CPUs, and directly increasing the number of machine nodes in the cluster version. It will have an immediate effect on query performance.

About the browser compatibility of Sensors products

(Effective date: from August 2, 2022)

Dear Sensors Customer:

Thank you for choosing to subscribe to Sensors products. Sensors products need to run on the PC through the browser. Different browsers have different rendering engines, and some browsers may have compatibility issues when running the product.

On June 15, 2022, Microsoft officially announced the end of support for Internet Explorer (hereinafter referred to as "IE") browser (click to view details). IE browser will no longer be upgraded or maintained in the future, which will inevitably cause compatibility issues for IE. Therefore, Sensors will no longer adapt or make any compatibility work for IE, while after rigorous testing, compatibility issues affecting page display may also exist in QQ, WeChat, Sogou, and other browsers.

Therefore, in order to ensure your experience when using Sensors products, it is recommended that you use the following browsers to operate Sensors products:

• Apple Safari (latest version)
• Google Chrome (latest version)
• Mozilla Firefox (latest version)
• Microsoft Edge (latest version)

You can refer to the following table for detailed information on Sensors product compatibility with different browsers:

In addition, SensData products also have related reminders. When your current browser version is too low or does not support access to SensData products, the system will automatically remind you to upgrade or install a compatible browser.

Finally, thank you for your continued support of Sensors Data. If you have any issues with the browser compatibility of SensData products, please contact your Customer Success Manager or Project Manager for feedback. We hope that through our services, SensData products can fully demonstrate their value to your business!

Sensors Data

August 2, 2022