シイエヌエス List of Products and Services

A first-year working professional with no practical experience in OCI (majoring in agriculture at university) introduces their study methods for obtaining the "OCI Architect Associate" certification. The OCI Architect Associate is one of the Oracle Cloud Infrastructure (OCI) certifications and is of intermediate difficulty. The exam categories are divided as follows: compute 20%, network 35%, storage 25%, and identity and access management (IAM) 20%. The total study time is approximately 60 hours, with about 30 hours spent on watching videos, 20 hours on practice questions, and 10 hours on mock question sets. There is almost no difference between Become an OCI Architect Associate (2025) and 2024, and using only the My Learn materials is sufficient; with consistent study, one can definitely obtain the certification. *For more details about the blog, you can view it through the related link. Please feel free to contact us for further information.

This article summarizes the investigation of errors that cannot be avoided when operating Oracle DB, based on actual experiences in the field. When an error occurs, the first step is to check the alert logs. There are DB alert logs, ASM alert logs, and CRS alert logs. Next, we check the trace logs and incident logs. Using the example of ORA-04031, we introduce the basic methods for checking three key items in AWR. Additionally, we touch on the verification of Memory Dynamic Components and SQL ordered by Elapsed Time. Furthermore, we explain information that is useful for failure investigation, such as ASH (Active Session History) and TFA (Trace File Analyzer). *For more detailed content of the blog, please refer to the related links. For further inquiries, feel free to contact us.

This article explains how to conduct accuracy verification between multiple providers and models using LiteLLM. LiteLLM is a tool that allows you to make requests to over 100 various LLM providers using OpenAI's input and output formats, making it easy to send the same prompt to multiple models. In this instance, we requested source code generation to solve our unique FizzBuzz problem from three models: GPT-5.2 Codex, Claude Opus 4.5, and Gemini 3 Pro, and compared and evaluated their output results. We defined four evaluation criteria: task fidelity, tone and style, latency, and cost, and assessed the generated results from each model. By using LiteLLM, we were able to easily conduct accuracy verification between different providers and models. *For more details about the blog, you can view it through the related links. Please feel free to contact us for more information.*

This article explains the step-by-step procedure for installing GO-Global on a Windows server environment in Oracle Cloud Infrastructure (OCI). GO-Global is designed to be installed simply by clicking buttons, but since all settings screens and error messages are displayed in English, it can be confusing for those who are not proficient in English. In this article, in addition to the basic precautions for introducing GO-Global, we also provide translations and supplementary explanations of the content displayed on the screens in English. The goals are to have GO-Global installed on the Windows server, to be able to access specified applications using GO-Global, and to add applications within GO-Global and confirm their operation. The article covers the installation procedures for both the trial and commercial versions, license settings, and operation confirmation. *For more detailed content of the blog, please refer to the related links. For further inquiries, feel free to contact us.

This article organizes the considerations regarding various parameters related to performance while progressing with the design and construction of PostgreSQL, addressing the concerns of "how far should we investigate?" and "what values should we decide on?" The purpose of this article is to clarify what aspects should be considered in the design and tuning during the detailed design phase and testing phase of the system from a performance perspective. It is not merely a list of parameters or an explanation of configuration values; rather, it summarizes why these parameters become subjects of design consideration and under what system characteristics or issues adjustments should be considered, providing insights that lead to practical decision-making. It explains the parameters that should be examined, categorized into SQL optimization, execution plans, statistical information, VACUUM, investigation and analysis, timeouts, and other performance-related categories. *For detailed content of the blog, please refer to the related links. For more information, feel free to contact us.*

This article is an output of the content learned by a mid-career employee who has been in charge of ServiceNow for about two months. ServiceNow is a cloud platform that allows for centralized management of internal operations. In many companies, operations are often fragmented by department, leading to delays in collaboration between departments. It is designed with the aim of facilitating collaboration and improving efficiency in such operations. It is utilized in various scenarios, such as onboarding new employees, handling applications related to employees' life events, and sharing and documenting trouble response histories in the sales department. When implementing ServiceNow, it is common to proceed through steps such as inventorying operations, selecting modules, and identifying and discussing related departments. Our company has been providing support for the implementation of ServiceNow in collaboration with clients, and we plan to continue supporting companies considering implementation in a partnership model in the future. *For more details about the blog, please refer to the related links. Feel free to contact us for more information.*

This article introduces the method of using Microsoft 365 Copilot to investigate the changes in configuration items between the existing version and the new version when I was responsible for upgrading the application server in my work. Since covering everything manually takes time, I attempted to utilize generative AI to investigate the changes. It discusses strategies for narrowing down the prompts to obtain the desired answers, as well as the challenges posed by hallucinations and output variability in generative AI. As countermeasures for output variability, I introduce strategies such as presenting the premises in the same format each time, clearly outlining the requests in bullet points, specifying the response format, and not asking for too much at once. I believe that using generative AI as an entry point for investigation, while leaving the final judgment, such as fact-checking the output results, to humans is the best approach. *For more details on the blog, you can view it through the related links. Please feel free to contact us for more information.*

This is a participation report for Oracle AI World 2025, held in Las Vegas from October 13 to 16, 2025. This event is organized by Oracle Corporation, and the name has changed from "Oracle Cloud World" to clearly indicate a focus on AI. During the keynote, it was discussed how AI is integrated throughout the technology stack under the theme "Oracle AI Powering Your Business," along with examples of AI applications in the healthcare sector presented by Larry Ellison. The design philosophy of OCI, emphasizing "maximum performance, minimum cost, and maximum security," was also highlighted. As next-generation cloud services, new resource names such as Oracle AI Database 26ai and Oracle Autonomous AI Database were introduced, along with enhancements to Zero Trust Packet Routing (ZPR) and new technologies like Oracle Acceleron. *For more detailed information, please refer to the related links. Feel free to contact us for further inquiries.*

This article is based on a lecture held at Doshisha University’s Kyotanabe Campus in collaboration with SAS Institute Japan, themed "Marketing × Optimization." While companies can utilize vast amounts of customer data, the increasing options for strategies and business rules make it difficult to make appropriate decisions using traditional methods that rely on intuition and experience. We introduce a method that uses predictive models based on machine learning to estimate the effects of various strategies and employs mathematical optimization to determine "which customers to target, with which strategies, and when," under constraints such as budget and frequency of customer contact. It emphasizes that defining the objective function, defining the optimization targets, and defining the constraints are crucial for organizing business requirements for optimization. *For more details about the blog, you can view it through the related links. Please feel free to contact us for more information.*

This article introduces the steps to set up a learning environment for those interested in PostgreSQL but who have never used it before. PostgreSQL is an open-source software (OSS) RDBMS, which has the advantage of not incurring licensing fees. In this article, we explain the steps to install PostgreSQL from source code on a Raspberry Pi, as well as the procedure for introducing the extension module pg_statsinfo. pg_statsinfo corresponds to Oracle's AWR report and collects statistical information as snapshots at specified intervals. Additionally, we include findings and investigations regarding points of interest during the pg_statsinfo setup, such as the role of the pg_statsinfo.target_server parameter and how to specify the monitored database. *For more detailed content of the blog, you can view it through the related links. Please feel free to contact us for more information.

This article introduces a method using Azure OpenAI services and Python to format (cleanse) long text data for analysis. It processes text data from surveys, product reviews, complaints, and emails to extract specified elements and convert them into numerical values or categories. The structure involves using three components: Excel files, Azure OpenAI, and Python. It explains the steps for creating a resource group in Azure, creating an AOAI resource, configuring the AOAI network, deploying a generative AI model, creating functions to call the generative AI model in Python, reading and processing the Excel file, and outputting the results. As a reference, the cost for processing approximately 1,000 data entries with gpt-4o-mini is 47 yen, which is deemed sufficient for data cleansing. *For detailed content of the blog, please refer to the related links. For more information, feel free to contact us.*

This article attempts to have a blog post written by generative AI, using prompts created by the generative AI itself. First, I requested ChatGPT to create the prompts, and then I executed the generated prompts as they were to produce the blog article. A prompt is an instruction to the generative AI, and it is important to clearly communicate what you want to generate. The method of adjusting prompts to improve the quality of the output is called "prompt tuning." Additionally, the article explains the issue of hallucination, where generative AI returns incorrect answers. As a technique to mitigate hallucination, it introduces RAG (Retrieval-Augmented Generation), which generates responses based on externally provided input, allowing for answers that incorporate the company's unique information. *For more details about the blog, you can view it through the related links. Please feel free to contact us for more information.*

This article is aimed at beginners in PowerShell who are looking to create scripts. It explains how to retrieve values entered in the console and how to obtain the current directory, using file copying as the theme. PowerShell is a scripting language and CLI tool developed by Microsoft, and it is included as a standard feature in Windows. Cmdlets are structured according to the [verb]-[noun] naming convention, making them intuitive and easy to understand for beginners. The explanation of the script is divided into four parts: variable declaration, directory retrieval, function definition, and confirmation of execution, along with calling the function. *For more detailed content of the blog, you can view it through the related links. Please feel free to contact us for more information.*

This article introduces the procedure for upgrading EKS using Terraform. EKS stands for "Amazon Elastic Kubernetes Service," which is a managed service that allows you to use Kubernetes on AWS. The overall flow of the upgrade consists of three steps: draining the nodes, modifying the tfvars file, and applying the tfvars file. During the node draining process, the nodes on EKS are stopped, and the Pods are evicted and terminated. If the Pods are not evicted and terminated before the upgrade, it may lead to a PodEvictionFailure, causing the upgrade to fail. The upgrade is achieved by changing the value of kubernetes_version in the tfvars file to the next higher version and applying it. *For more detailed information, please refer to the related links. Feel free to contact us for further inquiries.*

This is a report article on the Oracle CloudWorld 2024 (OCW) hosted by Oracle Corporation, which took place in September 2024. OCW is an event where the latest innovations in cloud technology related to Oracle Cloud Infrastructure (OCI) are announced. During the keynote session, Oracle CEO Safra A. Catz presented successful business cases utilizing OCI and AI. A notable topic was the announcement of "Dedicated Region 25," which has reduced the number of racks required for the customer-specific region "OCI Dedicated Region Cloud@Customer (DRCC)" to three racks. Representatives from Japanese companies such as Nomura Research Institute, Mitsubishi Fuso, KDDI, and Fujitsu gave presentations. In the Japan Session, Mr. Mizusawa from Oracle Japan introduced new announcements and features at CloudWorld. *For more detailed information, please refer to the related links. Feel free to contact us for further inquiries.*

This article introduces the steps and syntax for setting up Oracle Cloud Infrastructure Identity and Access Management (IAM) policies for those who have handled OCI resources but have never created IAM policies. IAM policies are a mechanism for managing who can access Oracle Cloud Infrastructure resources and controlling operational permissions. It explains the basic syntax, including elements such as subject (group specification), verb (the four verbs: inspect/read/use/manage), resource-type (resource type), location (compartment or tenancy), and conditions. It also describes the inheritance and attachment of IAM policies, using a hierarchical compartment structure as an example for the writing method. *For more details, you can view the blog through the related links. Please feel free to contact us for more information.*